Effect of UDP‐glucuronosyltransferase 1A1 activity on risk for developing Gilbert's syndrome

Abstract

Variations at the six nucleotides −3279 (T > G), −53 (A[TA]₆TAA > A[TA]₇TAA), 211 (G > A), 686 (C > A), 1091 (C > T), and 1456 (T > G) in the UDP‐glucuronosyltransferase 1A1 (UGT1A1) gene were determined in 178 Taiwanese patients with Gilbert's syndrome and in 200 healthy adults. Every subject was classified as a genotype depending on variation status of the six nucleotides in the UGT1A1 gene. The UGT1A1 activity for each genotype was calculated and then those genotypes were divided into 10 subgroups (Q1~Q10) according to their UGT1A1 activities, by using 10% as an interval. There were 24 genotypes observed, with UGT1A1 activity ranged 9%~100% of normal. There were two and six subjects with Gilbert's syndrome and none of healthy controls carrying genotypes in the Q1 and Q2 subgroups, respectively. The odds of developing Gilbert's syndrome were significantly higher for subjects carrying genotypes in the Q3, Q4, and Q5 subgroups than for those with genotype in the Q10 subgroup (odds ratios: 240.22, 59.80, and 14.67, respectively, P < .001 for each). Among the 178 patients of Gilbert's syndrome, serum bilirubin value was inversely correlated with UGT1A1 activity (r = −.306, P < .001). The sensitivity was 72.0% and the specificity was 90.5% by using UGT1A1 activity ≦40% of normal as the cut‐off point to distinguish between healthy subjects and patients of Gilbert's syndrome. Our results demonstrate that UGT1A1 activity is certainly a determinate for serum bilirubin value and UGT1A1 activity ≦40% of normal is a proper risk factor for the development of Gilbert's syndrome.

INTRODUCTION

The UDP‐glucuronosyltransferase 1A1 (UGT1A1), a 533‐amino acid protein encoded by the UGT1A1 gene located on chromosome 2q37.1, is the sole enzyme responsible for the glucuronidation of bilirubin in humans. Variant and mutant UGT1A1 genes can cause benign or fatal inherited types of unconjugated hyperbilirubinemia, namely Gilbert's syndrome, Crigler‐Najjar syndrome type 2 (CN‐2) and CN‐1. The activities of the UGT1A1 enzyme are approximately 30%, 10%, and 0% of normal in patients with Gilbert's syndrome, CN‐2 and CN‐1, respectively, while the corresponding serum bilirubin levels range from 17 to 103 μmol/L in patients with Gilbert's syndrome, 103 to 342 μmol/L in those with CN‐2 and 342 to 855 μmol/L in patients with CN‐1. The cDNA of the human UGT1 gene was cloned in 1991, and genetic defects in the UGT1 gene that cause CN‐1, CN‐2, and Gilbert's syndrome were first reported in 1992, 1993, and 1995, respectively. Subsequent studies revealed that CN syndrome is very rare, while the prevalence of Gilbert's syndrome is about 5% to 10%. A number of studies have investigated whether certain single nucleotide polymorphisms (SNPs) in the UGT1A1 gene in patients with Gilbert's syndrome differ among ethnic groups.

The occurrence of the A(TA)₆TAA > A(TA)₇TAA allele (UGT1A1*28, rs8175347) at nucleotide −53 in the UGT1A1 gene is relatively rare (15% vs 40%) and the variation rate within the coding region is much higher (25% vs 0.1%) in Asians than in Caucasians. The A(TA)₇TAA/A(TA)₇TAA variant in the promoter region is the sole cause of Gilbert's syndrome in Caucasians, while homozygous variation at nucleotide 211 (G > A) (p.Gly71Arg, UGT1A1*6, rs4148323), homozygous variations at nucleotide −3279 (T > G) (UGT1A1*60, rs4124874) and A(TA)₇TAA/A(TA)₇TAA are the main causes of Gilbert's syndrome in Asians. Interaction of coding region variations with the Gilbert‐type promoter abnormality of the UGT1A1 gene may cause moderate degrees of unconjugated hyperbilirubinaemia and may lead to neonatal kernicterus. Therefore, for Asians, the relationship between the UGT1A1 gene and Gilbert's syndrome is more complicated than that for Caucasians. Reduced UGT1A1 activity is known to cause or increase the risk of developing Gilbert's syndrome; therefore, evaluating UGT1A1 activity rather than analyzing SNPs of the UGT1A1 gene should provide more information about the mechanism underlying the development of Gilbert's syndrome.

When protein expression of CDNAs was detected in COS‐7 cells or in Heptoma cell lines Huh 7, UGT1A1 activities have been determined in subjects carrying a heterozygote for c.686 C > A/p.Pro229Gln (UGT1A1*27, rs35350960); A(TA)₇TAA/A(TA)₇TAA; A(TA)₈TAA/A(TA)₈TAA; a homozygote for c.211 G > A; a heterozygote for c.211 G > A; a homozygote for c.1456 T > G/p.Tyr486Asp (UGT1A1*7, rs34993780); double homozygote for c.211 G > A/c.1456 T > G; a homozygote for c.−3279 T > G and a homozygote for c.1091 C > T/p.Pro364Leu (UGT1A1*63, rs34946978), respectively. Those results might be good references to calculate UGT1A1 activities for the subjects carrying multiple SNPs of UGT1A1 gene.

The c.− 53 A(TA)₆TAA > A(TA)₇TAA was first found by Holland's scholars; the c. − 3279 T > G, c.211 G > A, c.686 C > A; and c.1456 T > G were first found by Japanese scholars; while the c.1091 C > T was first observed by my team. In our previous studies comparing variation status of the UGT1A1 gene between patients with Gilbert's syndrome and healthy controls, we found a total of six polymorphisms [c.−3279 T > G, c.−53 A(TA)₆TAA > A(TA)₇TAA, c.211 G > A, c.686 C > A, c.1091 C > T and c.1456 T > G]. However, UGT1A1 activity has never been concerned. In the present study, we wanted to see whether UGT1A1 activity is a risk factor for developing Gilbert's syndrome and how low the activity is proper to predict Gilbert's syndrome.

PATIENTS & METHODS

Subjects

Gilbert's syndrome was diagnosed on the criteria of serum total bilirubin ≧27.4 μmol/L, ratio of (unconjugated bilirubin)/(total bilirubin) ≧80%, no overt sign of hemolysis over a period greater than 4 months, normal liver function test and normal glucose‐6‐phosphate dehydrogenase (G6PD) activity in erythrocytes. Bilirubin values were determined in the morning following ≧8 hours of fasting and measurements were carried out at least twice within 6 months. Bias of bilirubin value for every subject was less than 10% and the higher (highest) value was taken into account in this study. Finally, 178 Taiwanese patients with Gilbert's syndrome (124 males) were identified from the gastroenterology clinic at the Cathay General Hospital, Taipei, Taiwan. Another 200 healthy Taiwanese adults (140 males), with normal liver and hematological test results and serum total bilirubin ≦25.7 μmol/L, were collected from subjects having physical examinations at the same hospital as the control group. Patients and healthy controls were aged 21 to 60 and 22 to 62 years, respectively. The bilirubin values were in range of 27.4 to 68.4 μmol/L and 8.6 to 25.7 μmol/L in patients and controls, respectively. Age and bilirubin value were matched between men and women in both groups, respectively, and gender distribution and age were matched between the two groups. Written informed consent was obtained from all participants and the study protocol was approved by the research committee of the Cathay General Hospital, Taipei, Taiwan (IRB No. CT‐9317 of Cathay General Hospital [CGH] for grant CGH‐MR‐9306), following the principles outlined in the Declaration of Helsinki for all human or animal experimental investigations.

Genetic analysis

Polymerase chain reaction‐restriction fragment length polymorphism (PCR‐RFLP) method was used to determine the six variants of the UGT1A1 gene in all subjects, as described in Table . The DNAs of wild type, heterozygote and homozygote for each of the six SNPs, have been found in Taiwanese by the sequencing methods, were performed as the positive controls in each assay of PCR‐RFLP.

Primers, restriction enzymes and results in the six polymorphisms of UGT1A1 gene determined by the PCR‐RFLP method

Position	Primer	Sequence	Restriction enzyme	Result, base pair
686	625F	5′‐GATCACATGACCTTCCTCCA‐3′	Bsr I	C: 124
C > A	625R	5′‐TCACCTCTCTCTGAAGGAAT‐3′	…	A: 80, 44
Promoter	(TA)F	5′‐TAACTTGGTGTATCGATTGGT‐3′	…	A(TA)6TAA: 77
A(TA)nTAA	(TA)R	5′CTTTGCTCCTGCCAGAGGTT3′	…	A(TA)7TAA: 79
–3279	U–3279F	5′‐ACCAAAGAACATTCTAACGG‐3′	Dra I	T: 180, 20
T > G	U–3279R	5′‐GTTCTCAAATTGCTTTGTTTA‐3′	…	G: 200
211	U1F1	5′‐AGATACTGTTGATCCCAGTG‐3′	Ava II	G: 128, 18
G > A	U211R	5′‐CTTCAAGGTGTAAAATGGTC‐3′	…	A: 146
1091	U4F2	5′‐GCCAACATATCCTACATTGC‐3′	Bcl I	C: 209
C > T	U1091R	5′‐GTGATAAAGGCACGGGTGAT‐3′	…	T: 190, 19
1456	U5F3	5′‐GTGGAGTTTGTGATGAGGCA‐3′	Ava II	T: 270
T > G	U5R1	5′‐GGAAATGACTAGGGAATGGT‐3′	…	G: 197, 73

Underline _, mutagenesis site; PCR‐RFLP, polymerase chain reaction‐restriction fragment length polymorphism; UGT, UDP‐glucuronosyltransferase.

UGT1A1 enzyme activity

As shown in Table , the enzyme activity values for 686 CA, −53 A(TA)₇TAA/ A(TA)₇TAA, −3279 GG, 1091 TT, 211 GA, 211 AA, and 1456 GG variants were obtained from previously published studies, while those for 686 AA, −53 A(TA)₆TAA/A(TA)₇TAA, −3279 TG, 1091 CT and 1456 TG variants were estimated by calculation (UGT1A1 activity for 686 AA = [activity for 686 CA]², while activity for A[TA]₆TAA/A[TA]₇TAA = [activity for A(TA)₇TAA/A(TA)₇TAA]^1/2 and so on). For the subjects carrying multiple SNPs in the UGT1A1 gene, estimated UGT1A1 enzyme‐activities were obtained by calculations. For example, for individuals carrying the genotype (686 CC/A[TA]₆TAA/A[TA]₆TAA/−3279 TG/1091 CC/211 GA/1456 TT), the UGT1A1 enzyme activity (46%) was calculated by multiplying 77% (activity in subjects carrying −3279 TG) by 60% (activity in subjects carrying 211 GA). However, for the subjects carrying multiple SNPs in close linkage, the UGT1A1 enzyme activity of the SNP with the lowest activity among those SNPs was estimated.

Variation status of the six polymorphisms and their UGT1A1 activities

Nucleotide	UGT1A1 activity, % of normal	Reference
686	CA	37	14
	AA	14	…
–53	(TA)6TAA > A(TA)7TAA	50	…
	A(TA)7TAA/A(TA)7TAA	25.5	7
–3279	TG	77	…
	GG	60	16
1091	CT	60	…
	TT	36	17
211	GA	60.2	8
	AA	32.2	8
1456	TG	27.6	…
	GG	7.6	8

Abbreviation: UDP‐glucuronosyltransferase. UGT1A1 activity (estimated value): 14% calculated by ([37%]²). UGT1A1 activity (estimated value): 50% calculated by ([25.5%]^1/2). UGT1A1 activity (estimated value): 77% calculated by ([60%]^1/2). UGT1A1 activity (estimated value): 60% calculated by ([36%]^1/2). UGT1A1 activity (estimated value): 27.6% calculated by ([7.6%]^1/2).

The UGT1A1 activity for each genotype was calculated and then those genotypes were divided into 10 subgroups according to their UGT1A1 activities: Q1 (≦10% of normal), Q2 (11%~20%), Q3 (21%~30%), Q4 (31%~40%), Q5 (41%~50%), Q6 (51%~60), Q7 (61%~70%), Q8 (71%~80%), Q9 (81%~90%) and Q10 (91~100%).

Statistical analysis

The odds ratio (OR) for the development of Gilbert's syndrome was assigned 1.0 for (number of patients in Q10)/(number of controls in Q10). The Mantel‐Haenszel chi‐square test was then used to calculate the OR and the 95% confidence interval (CI) for the subjects with genotypes in subgroups Q1 to Q9. Pearson's correlation and receiver‐operating characteristic (ROC) plot were used when appropriate. A P value <.05 was considered to indicate statistical significance. All statistical analyses were performed with the statistical package SPSS for Windows (Version 18.0, SPSS Inc., Chicago, Illinois).

RESULTS

The positive controls of wild type, heterozygote and homozygote for each of the six SNPs were accurate in each assay of PCR‐RFLP. Of the six SNPs, only c.1456 T > G was not seen in the control group. In the control group, both heterozygous and homozygous variations were observed for c.−3279 T > G and c.211 G > A, while only heterozygous variations were observed for c.−53 A(TA)₆TAA > A(TA)₇TAA, c.686 C > A and c.1091 C > T. However, in the patient group, all six SNPs were observed. In the patient group, both heterozygous and homozygous variations were seen for c.686 C > A, c.−53 A(TA)₆TAA > A(TA)₇TAA, c.−3279 T > G and c.211 G > A, while only heterozygous variations were observed for c.1091 C > T and c.1456 T > G. Distributions for c.−3279 T > G and c.211 G > A were in Hardy‐Weinberg equilibrium in the control group (P = .196 and .458, respectively) as well as in the patient group (P = .722 and .353, respectively). However, in the patient group, distributions for c.−53 A(TA)₆TAA > A(TA)₇TAA and c.686 C > A deviated from Hardy‐Weinberg equilibrium (P = .028 and .017, respectively) (n = 55, 75, and 48 for A[TA]₆TAA/A[TA]₆TAA, A[TA]₆TAA/A[TA]₇TAA and A[TA]₇TAA/A[TA]₇TAA and n = 144, 29, and 5 for 686 CC, 686 CA, and 686 AA, respectively, as shown in Table ).

The 24 genotypes and their UGT1A1 enzyme activities

Geno‐type	c.686 C > A	c.–53A(TA)6TAA > A(TA)7TAA	c.–3279T > G	c.1091C > T	c.211G > A	c.1456 T > G	UGT1A1 activity, % of normal	Patientsn	Controlsn
1	CC	6/6	TT	CC	GG	TT	100	2	46
2	CC	6/6	TG	CC	GG	TT	77	6	41
3	CC	6/6	GG	CC	GG	TT	60	3	5
4	CC	6/7	TG	CC	GG	TT	50	6	23
5	CC	6/7	GG	CC	GG	TT	50	10	13
6	CC	6/6	TT	CC	GA	TT	60	2	30
7	CC	6/6	TT	CC	AA	TT	32	18	4
8	CC	6/6	TG	CC	GA	TT	46	20	19
9	CA	6/7	TG	CC	GG	TT	37	2	4
10	CA	6/7	GG	CC	GG	TT	37	5	2
11	CC	6/7	TG	CC	GA	TT	30	39	5
12	CC	6/6	TG	CT	GG	TT	46	1	3
13	CC	6/6	TG	CT	GA	TT	28	0	2
14	CC	6/6	TT	CT	GG	TT	60	0	1
15	CA	6/7	TG	CC	GA	TT	22	9	1
16	CC	7/7	GG	CC	GG	TT	25.5	30	0
17	CA	7/7	GG	CC	GG	TT	25.5	12	0
18	AA	7/7	GG	CC	GG	TT	14	5	0
19	CC	6/7	GG	CT	GG	TT	30	3	1
20	CC	6/6	GG	CT	GG	TT	36	1	0
21	CC	7/7	GG	CT	GG	TT	15	1	0
22	CC	6/6	TT	CC	AA	TG	9	1	0
23	CA	6/7	TG	CC	GG	TG	10	1	0
24	CC	6/6	TG	CC	GG	TG	21	1	0

Abbreviation: UGT: UDP‐glucuronosyltransferase. 6/6: A(TA)₆TAA/A(TA)₆TAA. 6/7: A(TA)₆TAA/A(TA)₇TAA. 7/7: A(TA)₇TAA/A(TA)₇TAA.

Since c.686 C > A was observed not only being absolute association with −53 A(TA)₆TAA/A(TA)₇TAA or −53 A(TA)₇TAA/A(TA)₇TAA but also being absolute association with −3279 TG or −3279 GG and −53 A(TA)₆TAA > A(TA)₇TAA was observed being absolute association with −3279 TG or −3279 GG, the lowest model of UGT1A1 enzyme activity was estimated for those genotypes. For example, for individuals carrying the genotype (686 AA/−53 A[TA]₇TAA/A[TA]₇TAA/−3279 GG/1091 CC/211 GG/1456 TT), the UGT1A1 enzyme activity was estimated as 14% of normal (the enzyme activity of 686 AA, the lowest among 686 AA, −53 A[TA]₇TAA/A[TA]₇TAA [25.5%] and −3279 GG [60%]), while for individuals carrying the genotype (686 CA/−53 A[TA]₇TAA/A[TA]₇TAA/−3279 GG/1091 CC/211 GG/1456 TT), the UGT1A1 enzyme activity was estimated as 25.5% of normal (the enzyme activity of −53 A[TA]₇TAA/A[TA]₇TAA, the lowest among 686 CA [37%], −53 A[TA]₇TAA/A[TA]₇TAA and −3279 GG [60%]). Another example, for individuals carrying the genotype (686 CC/−53 A[TA]₇TAA/A[TA]₇TAA/−3279 GG/1091 CC/211 GG/1456 TT), the UGT1A1 enzyme activity was estimated as 25.5% of normal (the enzyme activity of −53 A[TA]₇TAA/A[TA]₇TAA, lower than that of −3279 GG [60%]). UGT1A1 activities for genotypes 4, 5, 9 to11, 15 to19, 21, and 23 were calculated by such manners, as shown in Table .

As listed in Table , there were 24 genotypes with estimated UGT1A1 activities ranging from 9% to 100% of normal. Genotypes 16 to18 and 20 to 24 were only found in the patient group, while genotypes 13 and 14 were only seen in the control group. Those 24 genotypes were divided into 10 subgroups (Q1~Q10) according to their UGT1A1 activities, as listed in Table . Table shows that two and six subjects of Gilbert's syndrome and none of controls carried genotypes in the Q1 and Q2 subgroups, respectively. The odds of developing Gilbert's syndrome were significantly higher for subjects carrying genotypes in the Q3 (240.22, P < .001), Q4 (59.80, P < .001) and Q5 (14.67, P < .001) than those carrying genotypes in the Q10, respectively. As seen in Figure , there was an inverse correlation between bilirubin level and UGT1A1 activity in the patient group (n = 178, r = −.306, P < .001). For drawing the ROC plot to distinguish between healthy adults and patients of Gilbert's syndrome, 10 cut‐off points by using 10% of UGT1A1 activity as an interval were utilized, as shown in Figure . Figure indicates that, to predict Gilbert's syndrome, the sum (162.5%) of sensitivity (72.0%) plus specificity (90.5%) for UGT1A1 activity ≦40% of normal as the cut‐off point was the largest one when it was compared with other nine sums using other cut‐off points to do so.

Odds ratios for the development of Gilbert's syndrome in the 10 subgroups carrying different UGT1A1 activities

Subgroups	UGT1A1 activity, % of normal	Genotypes	Patientsn	Controlsn	OR	95% CI	P value
Q1	≦10	22, 23	2	0	…	…	…
Q2	11∼20	18, 21	6	0	…	…	…
Q3	21∼30	11, 13, 15, 16, 17, 19, 24	94	9	240.22	49.87∼1157.25	<.001
Q4	31∼40	7, 9, 10, 20	26	10	59.80	12.16∼293.99	<.001
Q5	41∼50	4, 5, 8, 12	37	58	14.67	3.36∼64.11	<.001
Q6	51∼60	3, 6, 14	5	36	3.19	0.59∼17.43	.180
Q7	61∼70	…	0	0	…	…	…
Q8	71∼80	2	6	41	3.37	0.64∼17.61	.151
Q9	81∼90	…	0	0	…	…	…
Q10	91∼100	1	2	46	1.0	…	…

Calculated by the Mantel–Haenszel chi‐square test; CI, confidence interval; OR, odds ratio; UGT, UDP‐glucuronosyltransferase.

Correlation between bilirubin value and UGT1A1 activity among 178 patients with Gilbert's syndrome (calculated by Pearson's correlation test; ○, 1 point; ●, 2 points; ▵, 3 points; ▴, 4 points; □, 5 points; ■, 6 points; ⋄, 7 points; and ◆, 13 points)

The ROC plot for 10 cut‐off points by using 10% of UGT1A1 activity as an interval to distinguish between healthy adults and patients of Gilbert's syndrome (*40% of normal for the UGT1A1 activity. For the cut‐off points of 60% and 70% of normal, the sensitivity [95.5%] and specificity [56.5%] are the same. For the cut‐off points of 80% and 90% of normal, the sensitivity [99.0%] and specificity [43.5%] are the same)

DISCUSSION

Since 2000, we have published 14 studies on the relationship between variants in the UGT1A1 gene and Gilbert's syndrome (or neonatal hyperbilirubinemia). In our previous reports, the study subjects were usually divided into several groups, such as those carrying the wild type UGT1A1 gene, those carrying heterozygous variations, subjects with compound heterozygous variations, and those carrying homozygous and compound homozygous variations in the UGT1A1 gene, because we assumed that UGT1A1 activity gradually decreased as the number of variant alleles increased. However, as more SNPs in the UGT1A1 gene were found, we determined that our method of classifying study subjects was inaccurate. For example, UGT1A1 activities may differ considerably among subjects who carry compound heterozygous variations because some carry two while others can carry three or four SNPs.

Classifying subjects based on genotype is a more suitable method because genotypes represent variation of all SNPs in the UGT1A1 gene. However, in this study, we realized that we would encounter a number of difficulties if we focused only on the effect UGT1A1 genotypes might have on the risk for Gilbert's syndrome. For example, it would be a laborious process to precisely analyze the OR for each genotype since the number (n = 24) of genotypes is too much. In addition, the sample size is too small in some genotypes that would lead the effect of those genotypes on the risk for Gilbert's syndrome being neglected. Moreover, the UGT1A1 activity in some subjects carrying different genotypes is the same or similar. Grouping on the basis of UGT1A1 activity rather than on the basis of genotypes reduced the number of potential subgroups, thereby simplifying the analysis. As we know, the present report probably is the pioneering study to concern the six SNPs of c.−3279 T > G, c.−53 A(TA)₆TAA > A(TA)₇TAA, c.211 G > A, c.686 C > A, c.1091 C > T, and c.1456 T > G simultaneously in analyzing the relationship between UGT1A1 activity and Gilbert's syndrome.

Performing in COS‐7 cells, UGT1A1 activities have been determined in subjects carrying a homozygote for c.211 G > A and in subjects carrying a heterozygote for c.211 G > A being 32.2% and 60.2% of normal, respectively. If it is estimated by calculation as we did in this study, activity for 211 GA will be 56.7% of normal ([32.2% (activity for 211 AA)]^1/2), very closing to that (60.2% of normal) of the determined activity. Therefore, the manners we used in this study for calculating the estimated UGT1A1 activities seem available. UGT1A1 enzyme activities for c.686 C > A and for c.−53 A(TA)₆TAA > A(TA)₇TAA have been determined before c.686 C > A was found being not only absolute association with −53 A(TA)₆TAA/A(TA)₇TAA or − 53 A(TA)₇TAA/A(TA)₇TAA but also absolute association with −3279 TG or − 3279 GG (this manuscript) and c.−53 A(TA)₆TAA > A(TA)₇TAA was found being absolute association with −3279 TG or − 3279 GG (this manuscript). Since −53 A(TA)₇TAA/A(TA)₇TAA alone did not significantly reduce transcription of UGT1A1, the UGT1A1 enzyme activities of −53 A(TA)₇TAA/A(TA)₇TAA (25.5% of normal) and 686 CA (37% of normal) ever reported seem not be that of variant for a SNP (c.−53 A[TA]₆TAA > A[TA]₇TAA or c.686 C > A) but might be the values of combined effect from all the linked SNPs, respectively. Therefore, in this study, we calculate UGT1A1 enzyme activity for every subject carrying c.686 C > A/c.−53 A(TA)₆TAA > A(TA)₇TAA/c.−3279 T > G or c.−53 A(TA)₆TAA > A(TA)₇TAA/c.−3279 T > G as that of the SNP with lowest UGT1A1 activity among the linked SNPs an individual possesses.

From this study, we found low UGT1A1 activity (eg, ≦40% of normal) is an important risk factor for Gilbert's syndrome. In this study, two subjects (UGT1A1 activity = 9% and 10% of normal, respectively) of Gilbert's syndrome carried genotypes in the Q1 subgroup, six subjects of Gilbert's syndrome carried genotypes in the Q2 subgroup, while none of controls carried genotypes in the Q1 and Q2 subgroups. It indicates that the subjects carrying genotypes with UGT1A1 activity ranged 9% to 20% of normal would suffer from Gilbert's syndrome. However, in the control group, 19 subjects (9.5%) carried genotypes with UGT1A1 activity ranged 22% to 37% (22%, n = 1; 28%, n = 2; 30%, n = 6; 32%, n = 4 and 37%, n = 6, respectively). Those results indicate that subjects carrying the genotypes with UGT1A1 activity in between 22% and 37% of normal do not obligate suffer from Gilbert's syndrome.

We found a linearly inverse correlation between bilirubin level and UGT1A1 activity in patients with Gilbert's syndrome. In one of our previous studies, we observed that bilirubin level correlated well with the frequency of variants in the UGT1A1 gene in healthy adults. Those results demonstrate that UGT1A1 activity affects bilirubin levels in both healthy subjects and patients with Gilbert's syndrome.

We observed that genotypes 16 to 18 and 20 to 24 were only seen in the patient group. This is not surprising because UGT1A1 activities in subjects carrying those genotypes are low enough to cause Gilbert's syndrome (around 30% [36%] of normal for genotype 20 [n = 1] and around or below 20% of normal for the other seven genotypes [n = 51]). On the other hand, 28.1% (50/178) of subjects in the patient group carried genotypes with UGT1A1 activity ≧41% of normal and were assigned to subgroups Q5, Q6, Q8, and Q10. For those patients, SNPs other than the six variations in the UGT1A1 gene found in our general population may be observed. Although UGT1A1 is the sole enzyme responsible for the glucuronidation of bilirubin in humans, low UGT1A1 activity is not the sole risk factor for the development of Gilbert's syndrome. Indeed, in addition to defective bilirubin conjugation (low UGT1A1 activity), such as defective bilirubin uptake by hepatocytes (eg, polymorphisms in the organic anion transporting polypeptide 2, defective heme oxygenases, defective biliverdin reductase A, G6PD deficiency, and breast feeding) are also risk factors for the development of unconjugated hyperbilirubinemia. Those risk factors are worthy of investigation in patients suffering from hyperbilirubinemia with low to moderate UGT1A1 activity (>41% of normal).

Our results show that genotype distributions for c.686 C > A and c.–53 A(TA)₆TAA > A(TA)₇TAA in the patients with Gilbert's syndrome were not in Hardy‐Weinberg equilibrium. However, it is not surprising that variants of the UGT1A1 gene, for example, c.686 C > A and c.−53 A(TA)₆TAA > A(TA)₇TAA, are more prevalent in patients with unconjugated hyperbilirubinemia than in healthy subjects. It may be the reason, at least partially, why genotype distributions for those variants are not in Hardy‐Weinberg equilibrium. Nevertheless, those phenomena did not influence the results of this study.

Our results demonstrate that using UGT1A1 activity ≦40% of normal as the cut‐off point to distinguish between healthy subjects and patients of Gilbert's syndrome seems a good marker. Such a model to evaluate the association of UGT1A1 activity with metabolic rate for certain drugs or with risk for certain diseases can be applied to study other topics of personalized medicine, such as pharmacogenetics and cancer risk.

CONFLICT OF INTEREST

All authors declare no conflicts of interests.