A Gilbert's syndrome UGT1A1 variant confers susceptibility to ... - Nature

The Pharmacogenomics Journal (2004) 4, 49–53 & 2004 Nature Publishing Group All rights reserved 1470-269X/04 $25.00 www.nature.com/tpj

ORIGINAL ARTICLE

A Gilbert’s syndrome UGT1A1 variant confers susceptibility to tranilast-induced hyperbilirubinemia TM Danoff1 DA Campbell2 LC McCarthy3 KF Lewis3 MH Repasch1 AM Saunders4 NK Spurr4 IJ Purvis3 AD Roses4 C-F Xu3 1 Clinical Pharmacology and Discovery Medicine, GlaxoSmithKline, Philadelphia, PA, USA; 2 Etiologics Ltd., Medical Research Council, Harwell, Oxfordshire, UK; 3Discovery Genetics, GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage, UK; 4Genetics Research, GlaxoSmithKline, Research Triangle Park, NC, USA

Correspondence: C-F Xu, Discovery Genetics, GlaxoSmithKline Medicines Research Centre, Gunnels Wood Road, Stevenage SG1 2NY, UK. Tel: þ 44 1438 76 8392 Fax: þ 44 1438 76 4231 E-mail: [email protected]

Received: 31 July 2003 Revised: 26 September 2003 Accepted: 02 October 2003 Published online 2 December 2003

ABSTRACT

Tranilast (N-(30 40 -demethoxycinnamoyl)-anthranilic acid (N-5)) is an investigational drug for the prevention of restenosis following percutaneous transluminal coronary revascularization. An increase in bilirubin levels was observed in 12% of patients upon administration of tranilast in a phase III clinical trial. To identify the potential genetic factors that may account for the drug-induced hyperbilirubinemia, we examined polymorphisms in the uridine diphosphate glucuronosyltransferase 1A1 (UGT1A1) gene in over a thousand patients. Our results suggested that the TA repeat polymorphism in UGT1A1, which predisposes some individuals to Gilbert’s syndrome, predicted the susceptibility to tranilast-induced hyperbilirubinemia. The (TA)7/(TA)7 genotype was present in 39% of the 127 hyperbilirubinemic patients vs 7% of the 909 controls (P ¼ 2 1022). Rapid identification of genetic factors accounting for the observed adverse effect during the course of a double-blind clinical trial demonstrated the potential application of pharmacogenetics in the clinical development of safe and effective medicines. The Pharmacogenomics Journal (2004) 4, 49–53. doi:10.1038/sj.tpj.6500221 Published online 2 December 2003 Keywords: Gilbert’s syndrome; pharmacogenetics; tranilast; UGT1A1; hyperbilirubinemia

INTRODUCTION Percutaneous transluminal coronary revascularization (PTCR) is widely used for the treatment of ischemic heart disease; however, restenosis requiring repeat procedures remains problematic. In the 6 months following successful PTCR, restenosis may occur in up to 50% of patients.1 Tranilast (N-(30 40 -demethoxycinnamoyl)-anthranilic acid (N-5)) is an antiallergic compound, clinically effective in the control of autoimmune disease such as bronchial asthma and atoptic dermatitis. Based on previous clinical trials,2 tranilast was hypothesized to be effective in reducing the frequency of cardiovascular events and angiographic restenosis in patients undergoing PTCR. The PRESTO (Prevention of REStenosis with Tranilast and its Outcomes) study was designed to determine the safety and efficacy of tranilast in the prevention of restenosis after PTCR in Western populations.3 During this phase III clinical trial, hyperbilirubinemia, characterized by raised unconjugated serum bilirubin levels, was observed in approximately 12% of patients. The etiology of this mild elevation in serum

Genetic susceptibility to tranilast-induced hyperbilirubinemia TM Danoff et al

50

bilirubin level was unknown, but raised risks of major hepatic complications associated with tranilast. Serum bilirubin is derived primarily from degradation of hemoglobin and transported to the liver by binding to albumin. Within the heptocytes, uridine diphosphate glucuronosyltransferase 1A1 (UGT1A1) is the only enzyme that contributes to bilirubin glucuronidation.4 Glucuronidation is essential for the biliary elimination of bilirubin, and deficiencies in the UGT1A1 gene result in the accumulation of unconjugated bilirubin in the serum. In humans, there are at least three forms of inheritable hyperbilirubinemia caused by deficiencies of UGT1A1. Crigler–Najjar syndrome type 1 is characterized by lethal hyperbilirubinemia,5 and is caused by complete lack of UGT1A1 activity.6 Left untreated, Crigler–Najjar syndrome type 1 results in death from bilirubin encephalopathy. Crigler–Najjar syndrome type 2 is differentiated by the intermediate levels of hyperbilirubinemia, and is the result of severe deficiencies in UGT1A1 activity (B10% of normal activities). The incidences of both syndromes are very rare.7,8 A third condition, Gilbert’s syndrome,9 is characterized by mild hyperbilirubinemia, and is a benign form of episodic jaundice that occurs in approximately 3–10% of the population.10,11 Gilbert’s syndrome is thought to be harmless and often goes undetected. Baseline bilirubin levels of Gilbert’s syndrome patients are generally within normal limits. Episodes of hyperbilirubinemia can be brought on by a number of triggers such as psychological stress, infection, fasting and physical activities.10,11 Hepatic UGT1A1 activity in individuals with Gilbert’s syndrome is approximately 30% that of normal healthy individuals.12 Several studies have demonstrated that the genetic predisposition to Gilbert’s syndrome is associated with a TA repeat expansion in the TATAA box of the UGT1A1 gene.13–15 Individuals homozygous for the (TA)7 allele have higher serum bilirubin levels when compared to individuals who carry one or more copy of the (TA)6 allele.13,15–17 Reporter construct analysis showed that the activity of the A(TA)7TAA promoter was approximately 18–33% that of the wild-type construct.13 Furthermore, in vivo experiments demonstrated that the hepatic bilirubin UGT activity was reduced in (TA)7 homozygotes.18 The frequency of the (TA)7 allele is 30–40% in Caucasians,13,15,17,19 while it is less frequent in the Japanese

Table 1

population, with the frequency being approximately 16%.20,21 It was suggested that the GLY71Arg polymorphism may contribute to Gilbert’s syndrome, particularly in Japanese and Asian patients.14,22,23 Given these data, we set out to investigate the possible association of both the TA repeat and the GLY71Arg polymorphisms within UGT1A1, with hyperbilirubinemia observed during repeat dosing with tranilast in the PRESTO cohort. Our results supported the hypothesis that genetic predisposition to Gilbert’s syndrome predicted susceptibility to hyperbilirubinemia following administration of tranilast. RESULTS Genotype results for the TA repeat polymorphism of UGT1A1 were generated for 1042 patients. The allele frequencies for (TA)5, (TA)6, (TA)7 and (TA)8 were 0.001, 0.67, 0.33 and 0.001, respectively. The (TA)5 allele appears to function normally, while the (TA)8 allele, similar to the (TA)7 allele, is associated with Gilbert’s syndrome.19 Six individuals who had either the (TA)5 or the (TA)8 allele were excluded due to their low frequencies, with 1036 individuals remaining in subsequent analysis. Significant differences in genotype distributions were found between cases (bilirubin levels 42.00 mg/dl) and controls (bilirubin levels p2.00 mg/dl) (Table 1, P ¼ 2 1022). In all, 39% of the 127 cases carried the (TA)7/(TA)7 genotype, in comparison with only 7% of the 909 controls. These data predicted an odds ratio of 8.2 for cases vs controls, in relation to (TA)7 homozygotes vs any other TA genotypes. Genotype frequencies of the TA repeat polymorphism between the same hyperbilirubinemia cases and the super controls (bilirubin levels p1.00 mg/dl) were also compared (Table 1, P ¼ 8 1033). Only 3% of the 541 super controls had the (TA)7/(TA)7 genotype. These data predicted an odds ratio of 22.8 for the cases vs the super controls, in relation to (TA)7 homozygotes vs any other TA genotypes. It is interesting to note that the level of significance increased from 1022 to 1033, even though the number of controls reduced from 909 to 541 in the super control subset, illustrating the impact of using ‘stringently’ instead of ‘loosely’ defined control groups on the power of association studies. These findings agreed with the observations discussed by Schork et al,24 who described the importance of the definition of the control subjects, and

Genotype distribution at the UGT1A1 TA repeat polymorphism in patients from a phase III clinical trial of tranilast

Hyperbilirubinemia status

(TA)6/(TA)6

(TA)6/(TA)7

(TA)7/(TA)7

Total

P-valuea

Cases Controls Super controlb Totalc

22 444 318 466

55 398 208 453

50 67 15 117

127 909 541 1036

2 1022 8 1033

a

(17%) (49%) (59%) (45%)

(43%) (44%) (38%) (44%)

(39%) (7%) (3%) (11%)

Genotype distribution comparison between cases (bilirubin level 42.00 mg/dl) and controls (bilirubin level p2.00 mg/dl) or super controls (bilirubin level p1.00 mg/dl). b Super control is a subset of the control set. c Data from six of the 1042 patients are not shown, who had either the (TA)5 or the (TA)8 allele.

The Pharmacogenomics Journal


51

Figure 2 Association between tranilast-induced increase in bilirubin levels and the TA repeat polymorphism of UGT1A1 in 1036 individuals from the PRESTO clinical trial.3 Mean and standard error are shown.

bilirubin level varied as a function of the TA repeat genotypes, with the (TA)7 allele being associated with higher levels of bilirubin level both pre- and post-tranilast administration. The average bilirubin level increased from 0.93 to 2.22 mg/dl in (TA)7 homozygotes, while it increased from 0.48 to 1.01 mg/dl in (TA)6 homozygotes and 0.58 to 1.31 mg/dl in heterozygotes. Figure 2 showed the increases in bilirubin levels according to genotypes of the TA repeat polymorphism. Four individuals were found to be heterozygous for the G/A substitution, giving rise to a frequency of 0.16% for the GLY71Arg change in this sample. These four individuals were homozygous for the (TA)6 allele for the TA repeat polymorphism. All four individuals were in the control group, suggesting that the GLY71Arg polymorphism was not associated with the observed hyperbilirubinemia after tranilast administration in this Caucasian sample.

Figure 1 Pre- and during tranilast administration serum bilirubin levels in 1036 patients from the PRESTO clinical trial.3 (a) cases (n ¼ 127, bilirubin levels 42.00 mg/dl); (b) controls (n ¼ 909, bilirubin levels p2.00 mg/dl); (c) super controls, which is a subset of controls (n ¼ 541, bilirubin levels p1.00 mg/dl).

how power decreases as less extreme control sample thresholds are used. Thus the analysis of the extremes (low and high) may increase the ability to detect sensitive and specific variants in smaller phase II clinical trials. The mean serum bilirubin level of the 1036 subjects was 0.5770.01 mg/dl before tranilast administration, and increased to 1.2870.03 mg/dl after tranilast. The average bilirubin level increased from 0.53 to 1.00 mg/dl in controls and from 0.46 to 0.74 mg/dl in super controls, and increased from 0.91 to 3.24 mg/dl in cases (Figure 1). The average

DISCUSSION Acute hyperbilirubinemia is a commonly reported side effect of a number of medicines.25–27 The presence of Gilbert’s mutation predisposes HIV-infected patients to indinavirinduced hyperbilirubinemia.25 Genetic variation of UGT1A1 has also been shown to be associated with both irinotecaninduced hyperbilirubinemia and increased drug toxicity.26,27 During our phase III trial of tranilast, hyperbilirubinemia was observed in up to 12% of patients.28 At the time, the etiology of this elevation in serum bilirubin level was unknown. The affected individuals showed no signs of liver damage or overt hemolysis; however, the observed reaction may have been indicative of severe adverse reaction to tranilast. As part of candidate gene variant experimental design, we studied genetic polymorphisms in UGT1A1 in these patients. Our data suggested that an underlying genetic predisposition to Gilbert’s syndrome accounts for a major proportion of genetic susceptibility to tranilastinduced hyperbilirubinemia. Variability in phenotype expression in patients with Gilbert’s mutation is well documented, with hyperbilirubinemia developing in patients with the lowest levels of hepatic

www.nature.com/tpj


52

UGT activity.18 It is known that individuals with Gilbert’s mutation have low levels of UGT1A1 activity,18 and tranilast inhibit UGT1A1 activity (data not shown). Thus, patients who are (TA)7 homozygotes manifest increased susceptibility to hyperbilirubinemia as a consequence of tranilastinduced inhibition of UGT1A1 activity. Variation in allele frequencies between ethnic groups has implications for worldwide development and surveillance of medicines, and is an important factor to consider when interpreting pharmacogenetic information. Frequencies of the (TA)6/(TA)6, (TA)6/(TA)7 and TA)7/(TA)7 genotypes were 45, 44 and 11%, respectively, in this Caucasian sample, with a calculated allele frequency for (TA)7 being 33%. This is in agreement with other reported frequencies in Caucasians.15,17,19 The frequency of the (TA)7 allele, however, is much lower in Japanese and Asian populations, resulting in 2– 3% frequency of (TA)7 homozygotes.19 This suggests that the proportion of patients manifesting hyperbilirubinemia after tranilast treatment may be correspondingly lower in Asians. Attrition of drugs during development due to safety concerns is one of the main problems in the pharmaceutical pipeline. This study showed that tranilast-induced hyperbilirubinemia susceptibility was partly due to the recipient of medicine carrying an allele known to be related to Gilbert’s syndrome. It was therefore possible to define a benign condition associated with tranilast-induced hyperbilirubinemia, rather than to suspect patients being susceptible to some unknown, potentially more severe hepatic complications. Rapid identification of an etiology associated with an adverse effect during the course of a Phase III clinical trial demonstrated the power and potential use of pharmacogenetics in clinical studies, especially since the study design is prospective and double blind. Such information is crucial in influencing the regulatory environment surrounding a medicine’s progress to commercialization. Unfortunately for the company, the development of tranilast was terminated due to the lack of efficacy in preventing restenosis of coronary vessels.27 If this drug had good efficacy and reached the marketplace, one would be able to identify the benign hyperbilirubinemic condition by monitoring bilirubin levels both before and during the course of the medicine. Genetic testing of DNA samples would not have been necessary in the management of this mild adverse effect. In summary, our study clearly demonstrates that genetic variation in UGT1A1 is associated with an increased risk of tranilast-induced hyperbilirubinemia. This genetic variation predisposes individuals to a benign condition, Gilbert’s syndrome. The administration of tranilast unmasked the hyperbilirubinemia condition in these patients. The rapid identification of genetic risk factors during the course of a clinical trial demonstrated the power and potential applications of pharmacogenetics in future development of safer medicines.

MATERIALS AND METHODS Samples Appropriate ethical committee/IRB approval was obtained to collect blood samples prospectively for genetic studies

The Pharmacogenomics Journal

from individuals participating in PRESTO, a phase III double-blind, placebo-controlled clinical trial of tranilast.3 Written informed consent to perform genetic analysis was obtained from 1042 Caucasians in the United States who had undergone PTCR for treatment of ischemic heart disease. GlaxoSmithKline Clinical Laboratories performed all clinical testing. The total bilirubin levels were measured before and after tranilast treatment for either 1 or 3 months. DNA was extracted using a Nucleon DNA extraction and purification kit (Tepnel Life Sciences, UK). Mild hyperbilirubinemia was defined quantitatively as total serum bilirubin levels greater than 2.00 mg/dl. Five individuals (0.5%) preadministration of tranilast and 127 (12%) individuals following administration of tranilast had bilirubin levels greater than 2.00 mg/dl. These 127 individuals were defined as cases in this study. Controls consisted of the remaining 909 subjects, whose total bilirubin levels were equal to or less than 2.00 mg/dl following administration of tranilast. A ‘super control’ subset of 541 patients was also defined, whose bilirubin level was less than 1.00 mg/dl post-tranilast administration. Single-Nucleotide Polymorphism Identification and Genotyping The TA repeat polymorphism (rs3064744) and the G/A substitution (rs4148323), which results in the GLY71Arg amino-acid change within exon 1 of UGT1A1 (MIM191740), were genotyped. The TA repeat polymorphism was genotyped by direct sequencing of PCR products. In brief, PCR primers (50 -GAGGTTCTGGAAGTACTTTGC-30 and 50 -CCAAGCATGCACACCAG-30 ) were designed using the sequence of UGT1A1 to amplify a 423 bp DNA fragment. PCR products were sequenced using the ABI Prism BigDye Terminator cycle sequencing chemistry according to the manufacturer’s instructions, and analyzed on an ABI 3700 capillary sequencer. The resultant forward and reverse sequence trace files were independently analyzed by two individuals and genotypes were assigned to each sample. The G/A substitution (rs4148323) was genotyped by TaqMans assay (Perkin-Elmer, USA). Primers (50 -CTAGCACCT GACGCCTCGTT-30 and 50 -CACAGGGTACGTCTTCAAGGT GTA-30 ) and probes (50 -ATCAGAGACGGAGCAT-30 (VIC) and 50 -ATCAGAGACAGAGCAT-30 (FAM)) were supplied by Invitrogen Life Technologies (UK) and Applied Biosystems (USA), respectively. An ABI 7900HT was used to read fluorescence signals for each sample. Statistical Analysis The association between genotypes and phenotype was determined using Fisher’s exact test. P-values were determined using the PROC FREQ procedure (http://www.sas. com). PROC FREQ computes exact P-values by generating R C tables. ACKNOWLEDGEMENTS We thank the significant contribution of our colleagues in Genetics Research, Clinical Pharmacology and Discovery Medicine. Thanks are also due to the tranilast clinical team, the principal investigators


53

and patients who participated in this study. This study was wholly funded by GSK Research and Development.

12

13

DUALITY OF INTEREST None declared 14

ABBREVIATIONS PRESTO PTCR SNP UGT1A1

Prevention of REStenosis with Tranilast and its Outcomes percutaneous transluminal coronary revascularization single-nucleotide polymorphism uridine diphosphate glucuronosyltransferase 1A1

15

16

17

REFERENCES 1 Nobuyoshi M, Kimura T, Nosaka H, Mioka S, Ueno K, Yokoi H et al. Restenosis after successful percutaneous transluminal coronary angioplasty: serial angiographic follow-up of 229 patients. J Am Coll Cardiol 1988; 12: 616–623. 2 Tamai H, Katoh O, Suzuki S, Fujii K, Aizawa T, Takase S et al. Impact of tranilast on restenosis after coronary angioplasty: tranilast restenosis following angioplasty trial (TREAT). Am Heart J 1999; 138: 968–975. 3 Holmes D, Fitzgerald P, Goldberg S, LaBlanche J, Lincoff AM, Savage M et al. The PRESTO (Prevention of restenosis with tranilast and its outcomes) protocol: a double-blind, placebo-controlled trial. Am Heart J 2000; 139: 23–31. 4 Bosma PJ, Seppen J, Goldhoorn B, Bakker C, Oude Elferink RP, Chowdhury JR et al. Bilirubin UDP-glucuronosyltransferase 1 is the only relevant bilirubin glucuronidating isoform in man. J Biol Chem 1994; 269: 17960–17964. 5 Crigler JF, Najjar VA. Congenital familial non-hemolytic jaundice with kernicterus. Pediatrics 1952; 10: 169–180. 6 Seppen J, Bosma PJ, Goldhoorn BG, Bakker CT, Chowdhury JR, Chowdhury NR et al. Discrimination between Crigler–Najjar type I and II by expression of mutant bilirubin uridine diphosphate-glucuronosyltransferase. J Clin Invest 1994; 94: 2385–2391. 7 Tukey RH, Strassburg CP. Human UDP-glucuronosyltransferases: metabolism, expression, and disease. Ann Rev Pharmacol Toxicol 2000; 40: 581–616. 8 Kadakol A, Ghosh SS, Sappal BS, Sharma G, Chowdhury JR, Chowdhury NR. Genetic lesions of bilirubin uridine-diphosphoglucuronate glucuronosyltransferase (UGT1A1) causing Crigler–Najjar and Gilbert syndromes: correlation of genotype to phenotype. Hum Mutat 2000; 16: 297–306. 9 Gilbert NA, Lereboullet P. La chole´mie simple familiale. Sein Me´d 1901; 21: 241–243. 10 Owens D, Evans J. Population studies on Gilbert’s syndrome. J Med Genet 1975; 12: 152–156. 11 Bailey A, Robinson D, Dawson AM. Does Gilbert’s disease exist? Lancet 1977; 1: 931–933.

18

19

20

21

22 23

24

25

26

27

28

Black M, Billing BH. Hepatic bilirubin UDP-glucuronyl transferase activity in liver disease and Gilbert’s syndrome. N Engl J Med 1969; 280: 1266–1271. Bosma PJ, Chowdhury JR, Bakker C, Gantla S, de Boer A, Oostra BA et al. The genetic basis of the reduced expression of bilirubin UDPglucuronosyltransferase 1 in Gilbert’s syndrome. N Engl J Med 1995; 333: 1171–1175. Aono S, Adachi Y, Uyama E, Yamada Y, Keino H, Nanno T et al. Analysis of genes for bilirubin UDP-glucuronosyltransferase in Gilbert’s syndrome. Lancet 1995; 345: 958–959. Monaghan G, Ryan M, Seddon R, Hume R, Burchell B. Genetic variation in bilirubin UPD-glucuronosyltransferase gene promoter and Gilbert’s syndrome. Lancet 1996; 347: 578–581. Borlak J, Thum T, Landt O, Erb K, Hermann R. Molecular diagnosis of a familial nonhemolytic hyperbilirubinemia (Gilbert’s syndrome) in healthy subjects. Hepatology 2000; 32: 792–795. Rauchschwalbe SK, Zuhlsdorf MT, Schuhly U, Kuhlmann J. Predicting the risk of sporadic elevated bilirubin levels and diagnosing Gilbert’s syndrome by genotyping UGT1A1*28 promoter polymorphism. Int J Clin Pharmacol Ther 2002; 40: 233–240. Raijmakers MT, Jansen PL, Steegers EA, Peters WH. Association of human liver bilirubin UDP-glucuronyltransferase activity with a polymorphism in the promoter region of the UGT1A1 gene. J Hepatology 2000; 33: 348–351. Beutler E, Gelbart T, Demina A. Racial variability in the UDPglucuronosyltransferase 1 (UGT1A1) promoter: a balanced polymorphism for regulation of bilirubin metabolism? Proc Natl Acad Sci USA 1998; 95: 8170–8174. Ando Y, Chida M, Nakayama K, Saka H, Kamataki T. The UGT1A1*28 allele is relatively rare in a Japanese population. Pharmacogenetics 1998; 8: 357–360. Soeda Y, Yamamoto K, Adachi Y, Hori T, Aono S, Koiwai O et al. Predicted homozygous mis-sense mutation in Gilbert’s syndrome. Lancet 1995; 346: 1494. Sato H, Adachi Y, Koiwai O. The genetic basis of Gilbert’s syndrome. Lancet 1996; 347: 557–558. Koiwai O, Nishizawa M, Hasada K, Aono S, Adachi Y, Mamiya N et al. Gilbert’s syndrome is caused by a heterozygous missense mutation in the gene for bilirubin UDP-glucuronosyltransferase. Hum Mol Genet 1995; 4: 1183–1186. Schork NJ. Power calculations for genetic association studies using estimated probability distributions. Am J Hum Genet 2002; 70: 1480–1489. Zucker SD, Qin X, Rouster SD, Yu F, Green RM, Keshavan P et al. Mechanism of indinavir-induced hyperbilirubinemia. Proc Natl Acad Sci USA 2001; 98: 12671–12676. Ando Y, Saka H, Ando M, Sawa T, Muro K, Ueoka H et al. Polymorphisms of UDP-glucuronosyltransferase gene and irinotecan toxicity: a pharmacogenetic analysis. Cancer Res 2000; 60: 6921–6926. Iyer L, Das S, Janisch L, Wen M, Ramirez J, Karrison T et al. UGT1A1*28 polymorphism as a determinant of irinotecan disposition and toxicity. Pharmacogenomics J 2002; 2: 43–47. Holmes Jr DR, Savage M, LaBlanche JM, Grip L, Serruys PW, Fitzgerald P et al. Results of Prevention of REStenosis with Tranilast and its Outcomes (PRESTO) trial. Circulation 2002; 106: 1243–1250.

www.nature.com/tpj