Dihydropyrimidine Dehydrogenase Deficiency, a Pharmacogenetic Syndrome Associated with Potentially Life-Threatening Toxicity Following 5-Fluorouracil Administration

Home , Dihydropyrimidinase, Thymidylate synthase, Uridine phosphorylase

CCC v4n3 9/7/04 5:40 PM Page 181

Comprehensive Review Dihydropyrimidine Dehydrogenase Deficiency, a Pharmacogenetic Syndrome Associated with Potentially Life-Threatening Toxicity Following 5-Fluorouracil Administration

Hany Ezzeldin, Robert Diasio

Abstract Dihydropyrimidine dehydrogenase (DPD) deficiency is a pharmacogenetic syndrome associated with potentially life- threatening toxicity following the administration of standard doses of 5-fluorouracil.This syndrome derives its importance from the fact that approximately 2 million patients receive the drug worldwide each year.Population studies have suggested that 4%-7% of the American population exhibit dose-limiting toxicity that might be associated with a genetic defect in the DPYD gene that encodes for the DPD enzyme. During the past several years it has become increasingly clear that genetics is a major determinant of the variability in drug response, accounting for the probability of drug efficacy and the likelihood of toxic drug reactions.This article briefly discusses the clinical presentation, laboratory diagnosis, pharmacokinetics, inheritance, and the clinical management options of DPD deficiency.The variability of DPD enzyme activity in population studies and the different DPYD alleles together with new phenotypic and genotypic methods of screening for DPD deficiency will also be reviewed.

Clinical Colorectal Cancer, Vol. 4, No. 3, 181-189, 2004 Key words: Adverse drug reactions, DPYD gene, Pharmacogenomics

Introduction displayed partial or profound dihydropyrimidine dehydroge- Adverse drug reactions account for more than 100,000 nase (DPD) deficiency3,4 as a result of a genetic defect. deaths in the United States per year and have been reported as Pharmacokinetic studies have shown that the catabolic path- the fourth leading cause of death after heart diseases, cancer, way of 5-FU metabolism is important in the systemic toxicity and stroke.1 Genetics is a major determinant of the variability and the antitumor efficacy of 5-FU–based treatment.5-7 Like in drug response, accounting for drug efficacy and toxicity. Al- many other cancer chemotherapy drugs, 5-FU has a relatively though 5-fluorouracil (5-FU) remains one of the most widely narrow therapeutic window, with a small difference between prescribed cancer chemotherapy drugs and is generally well the median therapeutic efficacious dose and the median toxic tolerated, a recent population study suggested that 31%-34% dose. Some examples of chemotherapy drugs with efficacy of treated patients exhibit dose-limiting toxicity.2 Approxi- and/or toxicity are known to be affected by gene expression mately 40%-50% of patients with grade 3/4 toxicity to 5-FU levels of enzymes involved in their metabolic pathway. Three examples are now being recognized in cancer chemotherapy and metabolizing enzymes associated with pharmacogenetic Division of Clinical Pharmacology, Department of Pharmacology syndromes: the DPD enzyme involved in the metabolism of 5- and Toxicology, Comprehensive Cancer Center, University of Alabama at Birmingham FU, 5-fluoro-2'-deoxyuridine, and capecitabine; the uridine Submitted: Apr 9, 2004; Revised: Jul 19, 2004; Accepted: Jul 19, 2004 diphosphate glucuronosyltransferase enzyme implicated in the metabolism of irinotecan; and the thiopurine methyltrans- Address for correspondence: Robert Diasio, MD, University of Alabama at Birmingham, Department of Pharmacology and Toxicology and UAB ferase enzyme responsible for the metabolism of 6-mercap- Comprehensive Cancer Center, 1824 6th Ave S, Wallace Tumor Insti- topurine, azathioprine, and 6-thioguanine. tute, Room 620, University of Alabama at Birmingham, Birmingham, The importance of DPD in 5-FU catabolism has been dra- AL 35294-3300 matically illustrated by a pharmacogenetic syndrome caused Fax: 205-975-5650; e-mail: [email protected] by molecular defects in the DPYD gene that can result in a

Electronic forwarding or copying is a violation of US and International Copyright Laws. Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by Cancer Information Group, ISSN #1533-0028, provided the appropriate fee is paid directly to Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923 USA 978-750-8400.

Clinical Colorectal Cancer September 2004 • 181 CCC v4n3 9/7/04 5:40 PM Page 182

Pharmacogenomics of DPD and Colorectal Cancer

Figure 1 Anabolic and Catabolic Pathways of 5-FU Metabolism5

Catabolic DPD Pathway DHFU FUPA FBAL CFBAL Mode of action of 5-FU Metabolic 85% of 5-FU Regulation of 5-FU Anabolic DNA dR-1-p TK PMPK PDPK polymerase Pathway FdUrd FdUMP FdUDP FdUTP 1) Incorporation into DNA 5-FU TP Competes with natural substrate 2) Inhibition of TS PRPP/OPRT dUMP for catalytic site on TS FUMP PMPK PDPK UK FUDP FUTP 3) Incorporation into RNA R-1-P RNA polymerase UP FUrd

The activity of the DPD enzyme regulates the bioavailability of 5-FU. The anabolic pathway is responsible for the incorporation of the active metabolites into RNA and DNA. The 5-FU generates FdUMP, which competes with deoxyuridine monophosphate (ie, FUMP; the natural substrate of thymidylate synthase), on the active catalytic site, preventing the formation of thymidylate, which is essential for DNA synthesis. The catabolic pathway is essentially responsible for approximately 85% of 5-FU metabolism. Abbreviations: CFBAL = carboxy-fluoro-β-alanine; DHFU = dihydrofluorouracil; dR-1-P = deoxyribose-1-phosphate; FBAL = α-fluoro-β-alanine; FdUDP = fluorodeoxyuridine diphosphate; FdUrd = fluorodeoxyuridine; FdUTP = fluorodeoxyuridine triphosphate; FUDP = fluorouridine diphosphate; FUMP = fluorouridine monophosphate; FUPA = fluoro-β-ureidopropionic acid; FUrd = fluorouridine; FUTP = fluorouridine triphosphate; OPRT = orotate phosphoribosyltransferase; PDPK = pyrimidine diphosphate kinase; PMPK = pyrimidine monophosphate kinase; PRPP = phosphoribosyl pyrophosphate; R-1-P = ribose-1-phosphate; TP = thymidine phosphorylase; TK = thymidine kinase; UK = uridine kinase; UP = uridine phosphorylase

complete (profound) or partial loss of DPD enzyme activity.8-11 the irreversible conversion of > 85% of administered 5-FU to Pharmacogenetics refers to the influence of genetic differ- the inactive metabolite 5-fluorodihydrouracil.5,19-21 This re- ences on patients’ response to drugs.12 duces the availability of 5-FU active metabolites, including 5- Most of the pharmacogenetic syndromes that were first iden- fluoro-2'-deoxyuridine monophosphate (FdUMP), which in- tified were monogenic, having a common allele or alleles re- hibits thymidylate synthase required by cells for the synthesis of sponsible for the variable response among individuals sharing thymidylate and consequently inhibits DNA synthesis.11 the same phenotype.13 However, in drugs that are metabolized by several different enzymes in a multistep metabolic pathway, Clinical Presentation the possibility of a genetic variation in any or all of the metabo- of Dihydropyrimidine lizing enzymes (polygenic) will result in a multimodal overlap- Dehydrogenase Deficiency ping frequency distribution, and the relatively simple one-to- The typical clinical presentation of DPD deficiency includes one relation observed, for example, with thiopurine methyl- fever (associated with neutropenia), mucositis, stomatitis, and transferase enzyme or the cytochrome P450 isoenzyme (ie, diarrhea,11,21-23 which is similar to that of an accidental over- CYP2D6) would not be observed.13 dose of 5-FU. Excessive availability of 5-FU metabolites11 af- Since the late 1980s, there has been an increasing number fect the primary target sites in both cases, leading to increased of case reports describing severe toxicity (including death) to side effects and toxicity.20 Nausea, vomiting, rectal bleeding, treatment with 5-FU resulting from DPD deficiency.3,11,14,15 and/or skin changes may also occur. Neurologic abnormalities To date, identification of DPD-deficient individuals has re- include cerebellar ataxia, a broad-based gait, and changes in quired determination of DPD enzyme activity. Although sev- cognitive function, and are often subtle, but when these symp- eral studies have concluded that DPD activity should be as- toms occur, there should be increased suspicion of DPD defi- sayed before 5-FU treatment, the enzyme assay is complicat- ciency. Changes in the level of consciousness, reaching a com- ed and not available as a screening test in most clinical treat- atose state, is mostly encountered in cases with profound DPD ment centers.16 Typically, DPD enzyme assays have been per- deficiency.2,11 Initially at the time of presentation, routine chem- formed after the occurrence of unanticipated toxicity. istry tests and urine analysis show no abnormalities secondary The pharmacogenetic syndrome DPD deficiency, associated to DPD deficiency. However, abnormal blood counts, with leu- with severe, potentially life-threatening toxicity after adminis- copenia, a decrease in the absolute neutrophil count (often near tration of 5-fluorouracil (5-FU) or a related drug,11 derives its 0), and possibly thrombocytopenia and anemia, are most typi- importance from the fact that approximately 2 million patients cally seen when 5-FU is administered as a bolus with or with- receive the drug worldwide each year,17and, unlike other phar- out leucovorin.11,21-23 macogenetic syndromes, it has been associated with fatality after administration of standard doses of 5-FU.18 Dihydropy- Laboratory Diagnosis rimidine dehydrogenase is the initial and rate-limiting step in of Dihydropyrimidine pyrimidine catabolism, being responsible for conversion of the Dehydrogenase Deficiency naturally occurring pyrimidines, uracil and thymine, to dihy- In general, the clinical diagnosis of DPD deficiency remains drouracil (UH2) and dihydrothymine (TH2), respectively (Fig- difficult because the appearance of life-threatening toxicity ure 1).5 Dihydropyrimidine dehydrogenase is responsible for secondary to treatment with 5-FU is typically the first symptom

182 • Clinical Colorectal Cancer September 2004 CCC v4n3 9/7/04 5:40 PM Page 183

Hany Ezzeldin, Robert Diasio

of this pharmacogenetic syndrome. Elevated levels of uracil Table 1 Sequence Variants in the DPYD Gene and/or thymine (the natural substrates for DPD) in plasma or Reported *DPYD Amino Acid urine were the basis for laboratory diagnosis of complete or Nucleotide Exon profound DPD deficiency.11,24 In the past, diagnostic laborato- DPD Activity Nomenclature Change Change ry methods have included mass spectrometry (mainly for con- DPYD*9A C29R 85T>C 2 firmation), thin-layer chromatography, and high-performance M166V 496A>G 6 liquid chromatography (HPLC) to assay for thymine and Sequence K259E 775A>G 8 uracil. Partial DPD deficiency is more difficult to detect with variations 11 the aforementioned methods. This has led to the development associated M406T 1217T>C of a radioenzymatic assay for the DPD enzyme.16 This radio- with normal DPYD*4 S534D 1601G>A 13 DPD activity metric assay remains the gold standard for diagnosing DPD de- DPYD*5 I543V 1627A>G 13 ficiency, even after the development of genotypic assays, be- DPYD*6 V732I 2194G>A 18 cause of the presence of the multiple heterozygous sequence variations detected in most DPD-deficient patients. Various P1023S 3067C>T 23 cells and tissues can be examined with this assay. Peripheral DPYD*12 R21Q 62G>A 2 blood mononuclear (PBM) cells have been particularly useful P86L 257C>T 4 because the DPD is present without the subsequent enzymes of M182K 545T>A 6 the pyrimidine catabolic pathway, making the assessment of DPD activity in PBM cells more suitable. S201R 601A>C 6 There have been many attempts to develop genotypic tests Y211C 632A>G 6 4,24-26 for the identification of DPD-deficient patients. The DPYD*8 R235W 703C>T 7 complexity of the DPYD gene (with 23 exons), combined Sequence DPYD*11 V335L 1003G>A 10 with the large number of reported sequence variations (> 30; variations 11 Table 1) and the fact that DPD enzyme assays are often not associated with DPYD*12 E386>Stop 1156G>T possible in patients who experience lethal toxicity, have re- decreased DPD S492L 1475C>T 12 activity sulted in limited usefulness of single-mutation genotyping DPYD*13 I560S 1679T>G 13 tests such as restriction fragment length polymorphism or sin- DPYD*2A Del e14 IVS14+1G>A 14 gle strand conformation polymorphism. In addition, the sequence variations reported in previous DPYD*3 Frame shift 1897delC 14 studies have not always been correlated with a specific phe- A777S 2329G>A 19 27-29 notype (eg, DPD enzyme activity). This has resulted in DPYD*9B R886H 2657G>A 21 confusion over which sequence variations result in decreased D949V 2846A>T 22 DPD activity. Earlier studies of DPD-deficient patients have reported homozygote mutations resulting in complete (pro- H978R 2933A>G 23 found) DPD deficiency.4,9,14,30,31 More recently, our laboratory DPYD*10 V995F 2983G>T 23 described the characterization of a profoundly DPD-deficient H25R 74G>A 2 patient having a complex compound heterozygote geno- DPYD*7 Frame shift Del TCAT295-298 4 type.24In addition to confirming a codominant pattern of in- Sequence heritance for this pharmacogenetic disease, this study demon- variations Frame shift 812 del T 8 with uncertain I370V 1108A>G 10 strated the utility of a familial approach to examine genotype or untested DPD and phenotype. However, many of the reported sequence vari- activity Frame shift Del TG 1039-1042 10 ants, as mentioned later in this review, are of questionable L572V 1714C>G 13 importance or impact on DPD enzyme activity. D974V 2921A>T 23 Pharmacokinetics DPYD*12 mutation is composed of the 2 sequence variants 62G>A, R21Q and of Dihydropyrimidine 1156G>T, E386>Stop. DPYD*9B mutation is composed of the 2 sequence variants Dehydrogenase Deficiency DPYD*9A and 2657G>A, R886H. The disposition and pharmacokinetics of 5-FU were evalu- ated in one of the first reported DPD-deficient patients, who logic half-life was 159 minutes, with a clearance of 70 had been shown to have no detectable DPD enzyme activity.11 mL/min/m2, and 89.7% of the administered H3-5-FU was ex- A test dose of 5-FU (25 mg/m2, 600 µCi [3H-6]-5-FU) was creted unchanged in the 24 hours after administration of the administered as an intravenous bolus, after which plasma, drug. This is in contrast to a group of 10 normal control pa- urine, and cerebrospinal fluid were sampled at specified tients who were given a typical dose of 5-FU (450 mg/m2), in times. The 5-FU was detected in the plasma at unusually high which the half-life was 13 minutes ± 7, clearance was 594 levels for ≥ 8 hours, demonstrating a markedly altered phar- mL/min/m2 ± 198, and 9.8% ± 1.6% of the administered macokinetic pattern for this patient. In this patient, the bio- 5-FU was excreted in the initial 24 hours.

Clinical Colorectal Cancer September 2004 • 183 CCC v4n3 9/7/04 5:40 PM Page 184

Pharmacogenomics of DPD and Colorectal Cancer

Inheritance of Dihydropyrimidine Hemodialysis and Hemoperfusion Dehydrogenase Deficiency Hemodialysis and hemoperfusion were suggested to rapid- The pattern of inheritance of DPD deficiency was investi- ly remove any drug remaining in the body of a patient who gated in a family study of one of the initial patients who de- has already finished receiving 5-FU and is suspected of hav- veloped severe toxicity after 5-FU administration and who ing a DPD deficiency. However, in the presence of normal was subsequently shown to have DPD deficiency.11 The pa- renal function, neither is likely to be effective, particularly be- tient’s father, daughter, and son had enzyme levels interme- cause 5-FU, even in the presence of complete DPD deficien- diate between the patient’s level (essentially 0) and the level cy, is rapidly cleared via the urine.11 observed in the normal population, suggesting that the pa- Alternative methods used to manage the increased risk of tient’s relatives were partially deficient as a result of a het- 5-FU toxicity include: erozygous mutant gene, whereas the patient may have inherited 2 copies of a mutant gene (homozygous). These data, to- Administration of the Pyrimidine Nucleosides. Administration gether with a history of consanguinity in this family, initial- of the pyrimidine nucleosides (thymidine or uridine) can be un- ly suggested an autosomal recessive pattern of inheritance. dertaken as soon as possible (within a few hours) after 5-FU Although earlier studies reported a single homozygous mu- administration to overcome the block in thymidylate synthesis tation in the DPYD gene associated with DPD deficien- resulting from the inhibition of thymidylate synthase by the 5- cy,4,26,27,32,33 subsequent reports demonstrated that multiple FU nucleotide FdUMP.23 This has been used in few patients, heterozygous mutations could occur in the same patient. A with questionable results. As with hemodialysis and hemoper- recent report from our laboratory that characterized a pro- fusion, this would have to be performed soon after the admin- foundly DPD-deficient patient and family members demon- istration of 5-FU. In addition, the unavailability of the thymi- strated that profound DPD deficiency can result from a com- dine and uridine nucleosides has made this option less feasible. plex compound heterozygote genotype.24 This same patient was previously thought to have an autosomal recessive pat- Administration of a Colony-Stimulating Growth Factor. Ad- tern of inheritance. ministration of a colony-stimulating growth factor (eg, granu- Also, a recent study of 3 patients who experienced lethal locyte colony-stimulating factor) for the granulocytic precur- toxicity after administration of a standard dose of 5-FU sors has been proposed15 because life-threatening toxicity showed that lethal toxicity can occur in partially DPD-defi- caused by DPD deficiency is often accompanied by febrile cient individuals (ie, heterozygote for a mutant gene) after neutropenia or agranulocytosis. Although this line of treatment administration of 5-FU and is not exclusive to profoundly has been used in the management of several cases, its benefit is 34 DPD-deficient individuals (ie, homozygous for a mutant still unproven and the timing of administration is likely critical. gene) as previously suggested.31 Therefore, population studies show that partial DPD deficiency is more common than Aggressive Supportive Care. Appropriate antibiotic cover- profound deficiency,25,32,35 and lethal toxicity to treatment age for bacterial infections as well as antiinfectious coverage with 5-FU may be more common than originally thought. for potential nonbacterial infections may be necessary. For The complicated heterozygote genotype seen in these pa- example, in cases of fungal infections, fluid and electrolyte tients combined with DPD deficiency being an autosomal support with hospitalization in the intensive care unit (eg, for codominant inherited syndrome preclude the use of simple sepsis with the additional need for hemodynamic support) genotypic assays that identify 1 or 2 mutations as a method should be used as necessary. for identifying DPD-deficient individuals. These multiple heterozygote genotypes (which are more difficult to accu- rately characterize) may be responsible for some of the con- Variability of Dihydropyrimidine flicting reports that suggested a lack of correlation between Dehydrogenase Enzyme Activity phenotype and genotype.28,29 in Population Studies To better understand the extent of DPD variability, studies Clinical Management were undertaken to test for DPD activity in different popula- of Dihydropyrimidine tions. It became apparent that this activity was variable even Dehydrogenase Deficiency within otherwise healthy “normal” individuals. In a population Currently, most cases of DPD deficiency are diagnosed study of 124 healthy individuals (45% men and 55% women), long after the administration of 5-FU because of the lack of a DPD activities followed a normal distribution, with an approx- user-friendly test that permits rapid screening for DPD defi- imate 6-fold range in DPD enzyme activity.36,37 Although low ciency. Despite the limited options in managing patients’ tox- DPD activity was observed in some female patients,37 no sig- icities, several approaches have been suggested as follows.15 nificant differences in DPD activity by sex or age were demonstrated in subsequent studies.38,39 These studies of healthy Termination of Further Administration of 5-FU adult populations37 and of cancer patient populations38,39 con- Termination of any further administration of 5-FU (or relat- sisting of men and women 20-70 years of age demonstrated no ed drug) if the clinical manifestation suggests DPD deficiency. obvious age-related variation in DPD activity.

184 • Clinical Colorectal Cancer September 2004 CCC v4n3 9/7/04 5:40 PM Page 185

Hany Ezzeldin, Robert Diasio

Although population studies of DPD activity have not been Figure 2 Three-Dimensional Structure of DPD Protein47 comprehensively performed in all racial groups, initial stud- C-ter ies in the United States suggested that there was no difference R21Q FS4-1028 FMN-1030 A777S V732F N -ter in DPD activity between white and black patients. However, H25R FS4-1029 L572V the frequency of DPD deficiency seems to vary among racial D949V 1560S groups, being much less common in Japanese and Chinese C29R S534D patients than in white and black patients. Population studies 1543V of healthy individuals permitted the establishment of a statis-

tical cutoff point for the normal enzyme activity with use of D995F the radioenzymatic assay described earlier. D949V M166V Human Dihydropyrimidine FS4-1027 H978R Dehydrogenase Enzyme P86L R886H In the late 1980s, when DPD deficiency was first reported or FS4-1026 suspected in patients with cancer, there was little knowledge of R235W 11,40 the human enzyme. The recognition of the increasingly im- E386>Stop portant role of DPD as a critical step in 5-FU metabolism and 1370V M182K pharmacology provided the impetus for undertaking the earlier V335L Y211C studies.41 Initial approaches focused on human liver because an- M406T S201R imal study results and clinical experience with 5-FU adminis- S492L tered intraperitoneally and by hepatic artery infusion suggested NDP-1032 FAD-1031 K259E that this catabolic enzyme was in abundance in the liver. Human

liver DPD was purified to homogeneity in a 5-step purification Stereo view of the DPD protein showing the different functional domains and their locations scheme that permitted characterization of the protein, including (FMN-1030; FAD-1031; NDP [NADPH]–1032; 4 iron sulfur motifs (FS4 [1026-1029]). Also shown is the location of 26 amino acid sequence variants of the total 32 sequence variants its substrate kinetics; the location of functional domains; and the reported in the literature. The 3-dimensional structure of the molecule offers a better view of preparation of a polyclonal antibody. The determination of the the proximity between the different amino acid residues and functional domains. This figure was produced using the free macromolecular Visualization software program N-terminal amino acid sequence in this protein contributed to (Protein Explorer) developed by Eric Martz.70 the elucidation of the complementary DNA (cDNA) sequence and the subsequent isolation and characterization of the DPD gene known as DPYD.19 Additionally, these studies have proven that 5-FU has a similar affinity to human DPD enzyme However, it does not replace recombinant studies that involve as the natural substrate uracil.19,42 expression systems that experimentally determine the effect of Further insight into the DPD protein was provided by the the different mutations on the enzyme activity. determination of the cDNA sequence, wherein various functional domains within the translated amino acid sequence DPYD Alleles could be predicted.43,44 Several potentially important regions The determination of the complete cDNA sequence and the within the linear sequence of amino acids were identified, in- use of fluorescence in situ hybridization has permitted the as- cluding flavin-binding regions, NADPH-binding regions, signment of the DPYD gene to the region of chromosome iron/sulfur domains, and the uracil-binding site. Awareness of 1p22.48 The DPYD gene consists of 23 exons with exon length these critical regions was made possible by studies of the con- that varies from 50 base pairs to > 1550 base pairs.49,50 More servation of the amino acid sequence during evolution of the recently, the promoter for the DPYD gene has been cloned DPD protein, which helped in predicting the possible func- with initial characterization, permitting examination of its pos- tional consequences for DPD activity resulting from various sible role in regulating expression of the DPYD gene through mutations in the coding region of the DPYD gene.45 its regulatory elements detected in the 5'UTR.51 Studies of the 3-dimensional structure of the protein were The cDNA from individuals with DPD deficiency was ex- initiated by the isolation of the pure protein in sufficient quan- amined to determine any obvious structural abnormalities that tities. These studies demonstrated that porcine DPD shares ap- could explain the apparent dysfunctional DPD protein. A G- proximately 99% homology with human DPD. The radi- to-A mutation in the 5' splicing recognition sequence of in- ographic diffraction studies have permitted the 3-dimensional tron 14 was the first mutation to be described in the DPYD modeling of the DPD protein dimer with its 2 identical gene. This mutation resulted in deletion of the 165 base pair polypeptide chains.46,47 Subsequently, software programs corresponding to exon 14,51 and, according to established were developed to illustrate the 3-dimensional structure of the nomenclature, it is known as DPYD*2A, (IVS14+1G>A).52 human DPD protein (Figure 2). The 3-dimensional structure Further population studies demonstrated that it is the most fre- of the DPD protein offered the possibility of predicting influ- quently detected mutation to be associated with DPD defi- ence of different mutations or sequence variations on the func- ciency.3,14,25,26,33,53 During the past 7 years, > 30 sequence tional domains of the protein, and consequently its activity. variants have been reported in the literature.4,24,28,30,32,33,35,54-

Clinical Colorectal Cancer September 2004 • 185 CCC v4n3 9/7/04 5:40 PM Page 186

Pharmacogenomics of DPD and Colorectal Cancer

Figure 3 Uracil Breath Test: Metabolism of [2-13C]-Uracil Figure 4 Impact of DPD Deficiency on Uracil Breath Test66 and Measurement of 13CO2 200 Orally administered Healthy Subjects dose of 13C-Uracil 180 Partially Deficient Subjects 160 Profoundly Deficient Subjects

*13C-Uracil 140 13 * C- Fluorouracil 120

100

Dihydropyrimidine NADP DOB dehydrogenase NADPH+ 80 60 *13C-Dihydrouracil 40 *13C-Dihydrofluorouracil 20 Dihydropyrimidinase 0 51015202530405060708090 100 110 120 130 140 150 160 170 180 *13C-B-Ureidopropionate Time (Minutes) *13C-a-Fluoro- B-Ureidopropionate

B-ureidopropionase A dose of 6 mg/Kg of 13C-uracil was given to normal and partially DPD-deficient individuals, and the percent change over time in uracil metabolism was measured by the 13C breath test. Delta over baseline is an expression of 13CO /12CO in the expired air. *13C- B-Alanine 2 2 *13C-a-Fluoro- B-Alanine

13 * CO2 into the potential functional consequence of sequence variant + NH3 can be obtained by determining whether the location of the amino acid change is a highly conserved area or a potentially 13 critical region of the molecule (eg, active site, critical binding Measurement of CO2 Interference Interference regions), which could be easily performed with a comparative filter filter amino acid sequence alignment of the protein in different for for species to examine the conservation of the critical functional 12CO 13CO 2 2 domains, binding sites, and important contact residues.45 The confusion in identifying the molecular basis for DPD deficiency in some individuals might be attributed to the complicated heterozygote genotype pattern for DPD deficiency Absorbance 13CO 2 detected in some patients. Family studies of normal and deficient members can provide valuable insight into the molecular basis of DPD deficiency and 5-FU toxicity.4

2400 2300 Wave number (cm-1) 2200 New Phenotypic Methods for Screening of Dihydropyrimidine Dehydrogenase Deficiency 62 Table 1 shows the exon location of many of these sequence The radioenzymatic assay described earlier has remained variants.63 Although many were initially thought to be alleles the gold standard for making the diagnosis of DPD deficien- with functional significance for DPD activity, it has become cy.16 This assay offers the capability of identifying individu- increasingly clear that many are polymorphisms with little or als who are completely deficient in DPD activity and those no obvious functional effect. In addition, if the frame shifts re- who are partially deficient; it can also differentiate within a ported in the literature truly occur, their quantitative impor- population with activity in the normal range (low normal, tance in DPD enzyme activity remains to be clarified in the mid-normal, and high normal). However, it is time-consum- population. In our own experience over 15 years with DPD de- ing and unsuitable for screening patients before the adminis- ficiency, we have not observed frame shifts in any patient with tration of 5-FU. More user-friendly assays were developed to DPD enzyme deficiency. determine plasma (or urine) levels of uracil or thymine.17,40 Expression systems were used to evaluate the effect of these Recently, monoclonal antibodies to DPD were devel- sequence variants on protein function (ie, activity). This ap- oped.64,65 These assays could typically permit diagnosis of proach includes the examination of the cDNA sequence con- completely DPD-deficient individuals; however, they are taining the sequence variant after formation of sufficient pro- somewhat less accurate in assessing partially DPD-deficient tein in an expression system to perform an assay of protein individuals (typically individuals with a heterozygous muta- function (ie, assessment of DPD activity).9 Additional insight tion), who represent the vast majority of DPD-deficient indi-

186 • Clinical Colorectal Cancer September 2004 CCC v4n3 9/7/04 5:40 PM Page 187

Hany Ezzeldin, Robert Diasio

viduals. In an alternative approach, a limited sampling model a result of the presence of excessive drug, prolonged exposure was developed to detect abnormalities of DPD metabolism and caused by the increase in elimination half-life, and the typi- 5-FU/5-dihydro-FU disposition in poor metabolizers. In this cally narrow therapeutic window for most cancer chemother- model, a 2–time point (45 and 180 minutes) assessment allowed apy drugs. Other drug-metabolizing enzymes, with detected the evaluation of drug exposure, suggesting that this approach sequence variants in coding and noncoding regions of their could potentially improve therapeutic drug monitoring for treat- genes, have been associated with abnormal patterns of drug ment optimization of 5-FU in patients with cancer.66 metabolism, altered drug pharmacokinetics, and severe, often Recently, a rapid (≤ 90 minutes), noninvasive, and cost-ef- life-threatening toxicity. fective breath test has been developed. This test is suitable for With the completion of sequencing of the human genome use in a clinical laboratory and permits the evaluation of DPD and the potential to rapidly screen individuals with use of activity (normal activity and partial or profound deficiency) powerful pharmacogenetic research tools, a new era in cancer before the administration of 5-FU.67 In this assay, an admin- chemotherapy is likely to emerge. Although pharmacogenet- 13 13 istered dose of stable isotope C-uracil is released as CO2 ics will remain important, there will be an increasing empha- after a 3-step catabolism (Figure 3). With this breath test, par- sis on pharmacogenomics, wherein the alteration in sequence tially DPD-deficient individuals have demonstrated very dif- or a difference in the expression of critical genes might pro- ferent breath patterns compared with those of individuals vide insights in future clinical therapeutic decisions.69 with normal DPD activity (Figure 4). References New Genotypic Methods for 1. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug re- Screening of Dihydropyrimidine actions in hospitalized patients: a meta-analysis of prospective stud- Dehydrogenase Deficiency ies. JAMA 1998; 279:1200-1205. 2. Toxicity of fluorouracil in patients with advanced colorectal cancer: Allele-Specific Polymerase Chain Reaction and effect of administration schedule and prognostic factors. Meta- Denaturing High-Performance Liquid Chromatography Analysis Group In Cancer. J Clin Oncol 1998; 16:3537-3541. Allele-specific polymerase chain reaction (ASPCR) and de- 3. Johnson MR, Diasio RB. Importance of dihydropyrimidine dehydrogenase (DPD) deficiency in patients exhibiting toxicity following naturing HPLC (DHPLC) are 2 new approaches that were de- treatment with 5-fluorouracil. Adv Enzyme Regul 2001; 41:151-157. veloped for the rapid detection of sequence variants in compli- 4. van Kuilenburg AB, Haasjes J, Richel DJ, et al. Clinical implica- cated genes like the DPYD gene. In ASPCR, specific primers tions of dihydropyrimidine dehydrogenase (DPD) deficiency in patients with severe 5-fluorouracil-associated toxicity: identification and fluorogenic probes are used to detect specific known muta- of new mutations in the DPD gene. Clin Cancer Res 2000; 6:4705- tions in a specific fragment of the gene, such as DPYD*2A.14 4712. However, this method is not suitable for the detection of un- 5. Heggie GD, Sommadossi JP, Cross DS, et al. Clinical pharmacokinetics of 5-fluorouracil and its metabolites in plasma, urine, and known mutations. Although DPYD*2A is thought to be the bile. Cancer Res 1987; 47:2203-2206. most common sequence variant, thereby justifying the develop- 6. Grem JL. 5-Fluoropyrimidines. In: Chabner BA, Long DL, eds. ment of a specific assay, multiple sequence variants are associ- Cancer Chemotherapy and Biotherapy. Vol. 2. Philadelphia: Lip- pincott-Raven, 1996:149-211. ated with DPD deficiency, which complicated the identification 7. Salonga D, Danenberg KD, Johnson M, et al. Colorectal tumors re- of the molecular basis of DPD deficiency. Therefore, it would sponding to 5-fluorouracil have low gene expression levels of dihy- be desirable to rapidly screen the entire coding region and po- dropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase. Clin Cancer Res 2000; 6:1322-1327. tentially additional regions responsible for regulating the ex- 8. Etienne MC, Lagrange JL, Dassonville O, et al. Population study of pression of DPD messenger RNA (eg, promoter). dihydropyrimidine dehydrogenase in cancer patients. J Clin Oncol Advantages of the DHPLC over other methods include the 1994; 12:2248-2253. 9. Wei X, McLeod HL, McMurrough J, et al. Molecular basis of the rapid examination of DNA fragments (< 1000 bp), for known human dihydropyrimidine dehydrogenase deficiency and 5-fluo- and unknown sequence variations, using programmed melt- rouracil toxicity. J Clin Invest 1996; 98:610-615. ing temperature and optimized buffer gradients. With use of 10. Lu Z, Zhang R, Carpenter JT, et al. Decreased dihydropyrimidine dehydrogenase activity in a population of patients with breast can- DHPLC, chromatographic patterns generated by DNA frag- cer: implication for 5-fluorouracil-based chemotherapy. Clin Can- ments different from normal wild-type DNA sequence can be cer Res 1998; 4:325-329. detected.68 By using this method, the entire coding sequence 11. Diasio RB, Beavers TL, Carpenter JT. Familial deficiency of dihydropyrimidine dehydrogenase. Biochemical basis for familial of the DPYD gene, including the promoter region, can be pyrimidinemia and severe 5-fluorouracil-induced toxicity. J Clin In- scanned within 12.5 hours. In GC-rich genes such as the vest 1988; 81:47-51. DPYD gene, usually characterized by the presence of multi- 12. Roses AD. Pharmacogenetics and the practice of medicine. Nature 2000; 405:857-865. ple melting domains, the use of multiple screening tempera- 13. Weinshilboum R. Inheritance and drug response. N Engl J Med tures ensures the detection of all sequence variations inherent 2003; 348:529-537. in a DNA fragment. 14. Johnson MR, Hageboutros A, Wang K, et al. Life-threatening toxicity in a dihydropyrimidine dehydrogenase-deficient patient after treatment with topical 5-fluorouracil. Clin Cancer Res 1999; 5:2006-2011. Conclusion 15. Morrison GB, Bastian A, Dela Rosa T, et al. Dihydropyrimidine de- Dihydropyrimidine dehydrogenase deficiency is an exam- hydrogenase deficiency: a pharmacogenetic defect causing severe adverse reactions to 5-fluorouracil-based chemotherapy. Oncol Nurs ple that illustrates the impact of pharmacogenetics on cancer Forum 1997; 24:83-88. chemotherapy. In DPD enzyme deficiency, death can occur as 16. Johnson MR, Yan J, Shao L, et al. Semi-automated radioassay for

Clinical Colorectal Cancer September 2004 • 187 CCC v4n3 9/7/04 5:40 PM Page 188

Pharmacogenomics of DPD and Colorectal Cancer

determination of dihydropyrimidine dehydrogenase (DPD) activity. 37. Milano G, Etienne MC, Cassuto-Viguier E, et al. Influence of sex and Screening cancer patients for DPD deficiency, a condition associat- age on fluorouracil clearance. J Clin Oncol 1992; 10:1171-1175. ed with 5-fluorouracil toxicity. J Chromatogr B Biomed Sci Appl 38. Fleming RA, Milano GA, Gaspard MH, et al. Dihydropyrimidine 1997; 696:183-191. dehydrogenase activity in cancer patients. Eur J Cancer 1993; 17. Scrip’s Cancer Chemotherapy Report. Scrip World Pharmaceutical 29A:740-744. News. London: PJB Publications Ltd, 2002. 39. Fleming RA, Milano G, Thyss A, et al. Correlation between dihy- 18. Diasio RB, Lu Z, Zhang R, et al. Fluoropyrimidine catabolism. In: dropyrimidine dehydrogenase activity in peripheral mononuclear Muggia FM, ed. Concepts, mechanisms, and new targets for chemo- cells and systemic clearance of fluorouracil in cancer patients. Can- therapy. Boston: Kluwer Academic Publishers, 1995:71-94. cer Res 1992; 52:2899-2902. 19. Lu ZH, Zhang R, Diasio RB. Purification and characterization of di- 40. Tuchman M, Stoeckeler JS, Kiang DT, et al. Familial pyrimidinemia hydropyrimidine dehydrogenase from human liver. J Biol Chem and pyrimidinuria associated with severe fluorouracil toxicity. N 1992; 267:17102-17109. Engl J Med 1985; 313:245-249. 20. Sweeny DJ, Diasio RB. Toxicity of antimetabolites. In: Powis G, 41. Diasio RB, Harris BE. Clinical pharmacology of 5-fluorouracil. Hacker MP, eds. Mechanisms of toxicity of anticancer drugs, a study Clin Pharmacokinet 1989; 16:215-237. in human toxicity. New York: Macmillan, 1991. 42. Lu ZH, Zhang R, Diasio RB. Comparison of dihydropyrimidine de- 21. Daher GC, Harris BE, Diasio RB. Metabolism of pyrimidine analogues hydrogenase from human, rat, pig and cow liver. Biochemical and and their nucleosides. The International Encyclopedia of Pharmacology immunological properties. Biochem Pharmacol 1993; 46:945-952. and Therapeutics. Vol. 1. Oxford: Pergamon Press, 1994. 43. Yokota H, Fernandez-Salguero P, Furuya H, et al. cDNA cloning 22. Harris BE, Carpenter JT, Diasio RB. Severe 5-fluorouracil toxicity and chromosome mapping of human dihydropyrimidine dehydroge- secondary to dihydropyrimidine dehydrogenase deficiency. A po- nase, an enzyme associated with 5-fluorouracil toxicity and con- tentially more common pharmacogenetic syndrome. Cancer 1991; genital thymine uraciluria. J Biol Chem 1994; 269:23192-23196. 68:499-501. 44. Albin N, Johnson MR, Diasio RB. cDNA cloning of bovine liver dihy- 23. Takimoto CH, Lu ZH, Zhang R, et al. Severe neurotoxicity following dropyrimidine dehydrogenase. DNA Sequence 1996; 6:231-238. 5-fluorouracil-based chemotherapy in a patient with dihydropyrimi- 45. Mattison LK, Johnson MR, Diasio RB. A comparative analysis of dine dehydrogenase deficiency. Clin Cancer Res 1996; 2:477-481. translated dihydropyrimidine dehydrogenase cDNA; conservation 24. Johnson MR, Wang K, Diasio RB. Profound dihydropyrimidine de- of functional domains and relevance to genetic polymorphisms. hydrogenase deficiency resulting from a novel compound heterozy- Pharmacogenetics 2002; 12:133-144. gote genotype. Clin Cancer Res 2002; 8:768-774. 46. Dobritzsch D, Schneider G, Schnackerz KD, et al. Crystal structure 25. Raida M, Schwabe W, Hausler P, et al. Prevalence of a common of dihydropyrimidine dehydrogenase, a major determinant of the point mutation in the dihydropyrimidine dehydrogenase (DPD) pharmacokinetics of the anti-cancer drug 5-fluorouracil. EMBO J gene within the 5'-splice donor site of intron 14 in patients with se- 2001; 20:650-660. vere 5-fluorouracil (5-FU)- related toxicity compared with controls. 47. Dobritzsch D, Ricagno S, Schneider G, et al. Crystal structure of the Clin Cancer Res 2001; 7:2832-2839. productive ternary complex of dihydropyrimidine dehydrogenase 26. Van Kuilenburg AB, Vreken P, Abeling NG, et al. Genotype and with NADPH and 5-ioduracil. Implications for mechanism of inhi- phenotype in patients with dihydropyrimidine dehydrogenase defi- bition and electron transfer. J Biol Chem 2002; 277:13155-13166. ciency. Hum Genet 1999; 104:1-9. 48. Takai S, Fernandez-Salguero P, Kimura S, et al. Assignment of the 27. Vreken P, Van Kuilenburg AB, Meinsma R, et al. Dihydropyrimi- human dihydropyrimidine dehydrogenase gene (DPYD) to chromo- dine dehydrogenase (DPD) deficiency: identification and expression some region 1p22 by fluorescence in situ hybridization. Genomics of missense mutations C29R, R886H and R235W. Hum Genet 1997; 1994; 24:613-614. 101:333-338. 49. Johnson MR, Wang K, Tillmanns S, et al. Structural organization of 28. Collie-Duguid ES, Etienne MC, Milano G, et al. Known variant the human dihydropyrimidine dehydrogenase gene. Cancer Res DPYD alleles do not explain DPD deficiency in cancer patients. 1997; 57:1660-1663. Pharmacogenetics 2000; 10:217-223. 50. Wei X, Elizondo G, Sapone A, et al. Characterization of the human di- 29. Fernandez-Salguero PM, Sapone A, Wei X, et al. Lack of correla- hydropyrimidine dehydrogenase gene. Genomics 1998; 51:391-400. tion between phenotype and genotype for the polymorphically ex- 51. Shestopal SA, Johnson MR, Diasio RB. Molecular cloning and pressed dihydropyrimidine dehydrogenase in a family of Pakistani characterization of the human dihydropyrimidine dehydrogenase origin. Pharmacogenetics 1997; 7:161-163. promoter. Biochim Biophys Acta 2000; 1494:162-169. 30. Vreken P, Van Kuilenburg AB, Meinsma R, et al. Identification of 52. McLeod HL, Collie-Duguid ES, Vreken P, et al. Nomenclature for novel point mutations in the dihydropyrimidine dehydrogenase human DPYD alleles. Pharmacogenetics 1998; 8:455-459. gene. J Inherit Metab Dis 1997; 20:335-338. 53. Gardiner SJ, Begg EJ, Robinson BA. The effect of dihydropyrimi- 31. van Kuilenburg AB, Muller EW, Haasjes J, et al. Lethal outcome of dine dehydrogenase deficiency on outcomes with fluorouracil. Ad- a patient with a complete dihydropyrimidine dehydrogenase (DPD) verse Drug React Toxicol Rev 2002; 21:1-16. deficiency after administration of 5-fluorouracil: frequency of the 54. van Kuilenburg AB, Dobritzsch D, Meinsma R, et al. Novel disease- common IVS14+1G>A mutation causing DPD deficiency. Clin causing mutations in the dihydropyrimidine dehydrogenase gene in- Cancer Res 2001; 7:1149-1153. terpreted by analysis of the three-dimensional protein structure. 32. Ridge SA, Sludden J, Brown O, et al. Dihydropyrimidine dehydro- Biochem J 2002; 364:157-163. genase pharmacogenetics in Caucasian subjects. Br J Clin Pharma- 55. Vreken P, van Kuilenburg AB, Meinsma R, et al. Dihydropyrimidine col 1998; 46:151-156. dehydrogenase deficiency. Identification of two novel mutations and 33. Kouwaki M, Hamajima N, Sumi S, et al. Identification of novel muta- expression of missense mutations in E. coli. Adv Exp Med Biol tions in the dihydropyrimidine dehydrogenase gene in a Japanese pa- 1998; 431:341-346. tient with 5-fluorouracil toxicity. Clin Cancer Res 1998; 4:2999-3004. 56. Meinsma R, Fernandez-Salguero P, Van Kuilenburg AB, et al. 34. Ezzeldin H, Johnson MR, Okamoto Y, et al. Denaturing high per- Human polymorphism in drug metabolism: mutation in the dihy- formance liquid chromatography analysis of the DPYD gene in pa- dropyrimidine dehydrogenase gene results in exon skipping and tients with lethal 5-fluorouracil toxicity. Clin Cancer Res 2003; thymine uracilurea. DNA Cell Biol 1995; 14:1-6. 9:3021-3028. 57. Gross E, Neubauer S, Seck K, et al. High-throughput mutation 35. Yamaguchi K, Arai Y, Kanda Y, et al. Germline mutation of dihy- screening of the DPYD gene by denaturing HPLC. Proc Am Assoc dropyrimidine dehydrogenese gene among a Japanese population in Cancer Res 2002; 43:322A. relation to toxicity to 5-Fluorouracil. Jpn J Cancer Res 2001; 58. Ridge SA, Sludden J, Wei X, et al. Dihydropyrimidine dehydroge- 92:337-342. nase pharmacogenetics in patients with colorectal cancer. Br J Can- 36. Lu Z, Zhang R, Diasio RB. Dihydropyrimidine dehydrogenase ac- cer 1998; 77:497-500. tivity in human peripheral blood mononuclear cells and liver: popu- 59. van Kuilenburg AB, Vreken P, Beex LV, et al. Severe 5-fluorouracil lation characteristics, newly identified deficient patients, and clini- toxicity caused by reduced dihydropyrimidine dehydrogenase activ- cal implication in 5-fluorouracil chemotherapy. Cancer Res 1993; ity due to heterozygosity for a G-->A point mutation. J Inherit 53:5433-5438. Metab Dis 1998; 21:280-284.

188 • Clinical Colorectal Cancer September 2004 CCC v4n3 9/7/04 5:40 PM Page 189

Hany Ezzeldin, Robert Diasio

60. Christensen E, Cezanne I, Kjaergaard S, et al. Clinical variability in acterization of a monoclonal antibody with cross-reactivity towards three Danish patients with dihydropyrimidine dehydrogenase defi- uracil and thymine, and its potential use in screening patients treat- ciency all homozygous for the same mutation. J Inherit Metab Dis ed with 5-fluorouracil for possible risks. Clin Chim Acta 2002; 1998; 21:272-275. 322:59-66. 61. Albin N, Johnson MR, Shahinian H, et al. Initial characterization of 66. Di Paolo A, Danesi R, Vannozzi F, et al. Limited sampling model for the molecular defect in human dihydropyrimidine dehydrogenase the analysis of 5-fluorouracil pharmacokinetics in adjuvant chemo- deficiency. Proc Am Assoc Cancer Res 1996; 36:211A. therapy for colorectal cancer. Clin Pharmacol Ther 2002; 72:627-637. 62. Vreken P, van Kuilenburg AB, Meinsma R, et al. Dihydropyrimidine 67. Mattison LK, Carpenter M, Wang YH, et al. Rapid identification of dehydrogenase deficiency: a novel mutation and expression of mis- dihydropyrimidine dehydrogenase (DPD) deficiency by a novel 2- sense mutations in E. coli. J Inherit Metab Dis 1998; 21:276-279. 13C-uracil breath test (UBT). Clin Cancer Res 2004; 10:2652-2658. 63. Ezzeldin H, Diasio RB. Pharmacogenetics and Cancer Chemotherapy. In: 68. Ezzeldin H, Okamoto Y, Johnson MR, et al. A high-throughput de- Budman DR, Calvert AH, Rowinsky EK, eds. Handbook of Anticancer naturing high-performance liquid chromatography method for the Drug Development. Philadephia: Lippincott Williams & Wilkins, 2003. identification of variant alleles associated with dihydropyrimidine 64. Diasio RB, Van Kuilenburg AB, Lu Z, et al. Determination of dihy- dehydrogenase deficiency. Anal Biochem 2002; 306:63-73. dropyrimidine dehydrogenase (DPD) in fibroblasts of a DPD defi- 69. Evans WE, Relling MV. Pharmacogenomics: translating functional cient pediatric patient and family members using a polyclonal anti- genomics into rational therapeutics. Science 1999; 286:487-491. body to human DPD. Adv Exp Med Biol 1994; 370:7-10. 70. Martz E. Protein Explorer: easy yet powerful macromolecular visu- 65. Honda T, Inagawa H, Fukushima M, et al. Development and char- alization. Trends Biochem Sci 2002; 27:107-109.

Clinical Colorectal Cancer September 2004 • 189