Editor—Crigler-Najjar syndrome types 1 and 2 (CN1 and CN2) are inherited as autosomal recessive conditions and are characterised by severe non-haemolytic unconjugated hyperbilirubinaemia. CN1 is the most severe form, in which a virtual absence of hepatic bilirubin-uridinediphosphoglucuronate glucuronosyltransferase (UGT1A1) (PIR Accession A31340) activity results in serum bilirubin levels of 340-685 μmol/l or higher (normal 1.7-17.1 μmol/l).1 A milder variant of this disorder, termed CN2, is associated with intermediate levels of hyperbilirubinaemia (120-340 μmol/l) as a result of an incomplete deficiency of hepatic UGT1A1 activity. Usually, the residual UGT1A1 activity can be induced by phenobarbital administration, resulting in partial amelioration of jaundice.2 A third type of inherited unconjugated hyperbilirubinaemia, termed Gilbert's syndrome, is associated with mild, fluctuating hyperbilirubinaemia, ranging from normal levels to up to 85 μmol/l.3 Hepatic UGT1A1 activity is reduced to approximately 30% of normal in Gilbert's syndrome. In all three forms of inherited hyperbilirubinaemia, other hepatic functions are unaffected and the liver is morphologically normal.4 Before the institution of routine phototherapy, CN1 used to be uniformly lethal because of bilirubin encephalopathy (kernicterus).5 6 Although phototherapy has prolonged survival, it becomes progressively ineffective around puberty.6 7 Hepatocyte transplantation has resulted in partial amelioration of jaundice,8 9 but at present liver transplantation remains the only definitive therapy.10 11Kernicterus is much less common in CN2, and does not occur in Gilbert's syndrome.

UGT1A1 is one of the several UGT isoforms that are expressed from theUGT1A gene locus,12 located in human chromosome 2q37.13 The locus contains four exons at the 3′ end (exons 2, 3, 4, and 5), which are used in several UGT1A isoform mRNAs and encode the identical carboxy-terminal half of these enzymes. Upstream to these common region exons are a series of unique exons (exon 1A1 to 1A12), only one of which is used in a given UGT1A isoform mRNA. Each unique exon encodes the variable N-terminal region of one UGT1A isoform, which imparts its substrate specificity. The individual isoforms of the UGT1A subfamily are named according to the unique exon used. For example, UGT1A1 is encoded by exons 1A1, 2, 3, 4, and 5, and the gene is termed UGT1A1(GenBank accession No AF 180372, OMIM 191740). Following the initial report in 1992,14 studies from several laboratories have shown that CN1 and CN2 are caused by genetic lesions of any of the five exons of the UGT1A1 gene, or their flanking splice junctions. These mutations have been reviewed recently.15 16 The severity of the functional deficiency of UGT1A1 is determined by the nature of the genetic lesion.17 CN1 results from genetic lesions that cause premature truncation of UGT1A1 or substitution of critical amino acid residues, whereas CN2 is caused by substitution of single amino acid residues that markedly reduce, but do not abolish the catalytic activity of the enzyme. Although the incidence of Crigler-Najjar syndromes is not known precisely, these disorders probably occur in less than one in a million live births in the western hemisphere.11 Thus, the gene frequency can be estimated to be 1:1000 or less. Because the expression of one structurally normal allele is sufficient to keep serum bilirubin levels within normal limits, heterozygous carriers are not expected to have hyperbilirubinaemia.

In contrast to CN1 and CN2, Gilbert's syndrome is one of the most common inherited disorders in humans. In western populations, this condition is associated with a TA insertion within a dinucleotide repeat in the TATAA element, upstream of the first exon ofUGT1A1. The normal TATAA element sequence is A(TA)₆TAA, whereas in Gilbert syndrome the sequence is A(TA)₇TAA.18 The variant TATAA element reduces the expression of the structurally normal enzyme.18 Two additional polymorphisms of the TATAA element, a shorter (A(TA)₅TAA) and a longer (A(TA)₈TAA) variant, have been described in people of African origin.19 An inverse relationship between the length of the TATAA element and UGT1A1 expression has been reported. The common Gilbert-type TATAA element (A(TA)₇TAA) is present in homozygous state in approximately 9% of the European and North American population, the gene frequency being 30%.18 20-22 Homozygosity for the variant TATAA element is required, but not sufficient for the manifestation of the Gilbert phenotype, which is found in 4-7% of the general population.18 Other factors contribute to the serum bilirubin levels. For example, the Gilbert phenotype is seen less commonly in women, probably because of lower daily bilirubin production.23 In addition to the promoter abnormality, Japanese investigators have reported that some structural mutations ofUGT1A1 can result in mild hyperbilirubinaemia, compatible with the diagnosis of Gilbert's syndrome.24-26 These mutations have not been reported in white, black, south Asian, or Middle Eastern populations.

Interestingly, intermediate levels of hyperbilirubinaemia are observed commonly in families of patients with CN1 or CN2, which had given rise to conflicting opinions regarding the mode of inheritance of CN2 in the past.2 7 27 Because the gene frequency of the Gilbert-type promoter is 30% in the western population, a significant percentage of heterozygous carriers of CN-type mutations of the structural region of UGT1A1 would be expected also to carry a Gilbert-type promoter on one or both alleles. Sporadic cases of coexistence of a Gilbert-type promoter and a structural mutation of UGT1A1 have been reported.28-30 We postulated that such compound heterozygosity should be a more common cause of intermediate levels of jaundice than homozygosity for a structural mutation. We have examined this postulate in a systematic study over the course of the last three years by performing routine genetic analysis of a large number of referred patients with various levels of unconjugated hyperbilirubinaemia and their families. Here we report eight patients from four families, who had intermediate levels of unconjugated hyperbilirubinaemia because of compound heterozygosity of a Gilbert-type promoter and a structural region mutation ofUGT1A1. Three of the structural mutations are being reported here for the first time. In one of these families, a pair of dizygotic twins, who also had ABO incompatibility with the mother, developed severe neonatal hyperbilirubinaemia, leading to chronic encephalopathy.

Four families, designated families A to D, were studied. The probands in all the families showed clinical jaundice with predominantly unconjugated hyperbilirubinaemia. In families A, B, and D the children were the probands, while in family C the mother was the primary subject of the study. In all probands and their family members, the serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase levels were within normal limits and there was no evidence of any other liver disease or chronic haemolysis. Therefore, an inherited disorder of bilirubin glucuronidation was suspected.

Blood samples were obtained from all the subjects studied with appropriate informed consent. Genomic DNA was extracted from whole anticoagulated blood using the Qiagen kit (Qiagen, Valenica, CA), according to the manufacturer's instructions, and was used as template for amplification by polymerase chain reaction (PCR). A segment of theUGT1A1 upstream untranslated region, the five exons, and the flanking intronic sequences were amplified in three pieces using specific amplimers. A 104 bp segment upstream of the translation initiation codon (ATG), exon 1, and a short segment of intron 1 were amplified as a single amplicon. Exons 2, 3, and 4 and introns 2, 3, and a short segment of intron 4 were amplified as a second amplicon. Exon 5 and its flanking intronic sequences were amplified as the third amplicon. The amplicons generated by the first PCR were isolated by electrophoresis on low melting agarose gels and reamplified. The sequences and the locations of the primers used for the PCR reactions are as follows. Exon 1 (including 29 bp untranslated region + 85 bp upstream genomic sequence: sense: 5′ GTCACGTGACACAGT CAAAC 3′ (85 nt upstream of 5′ end of exon 1), antisense: 5′ GCTTGCTCAGCATATATCTGG 3′ (74 bp downstream of the 3′ end of exon 1); exons 2, 3, and 4, intron 2 and 3 and flanking DNA sequences: sense: 5′ CTCTATCT CAAACACGCATGCC 3′ (105 nt upstream of 5′ end of exon 2), antisense: 5′ TTATCATGAATGCCATGACC 3′ (29 nt downstream of 3′ end of exon 4); exon 5 and flanking DNA sequences: sense: 5′ GAGGATTGTTCATAC CACAGG 3′ (35 nt upstream of 5′ end of exon 5), antisense: 5′ TGAATTTAACACTGATTCTGTT 3′ (nt 1666 to nt 1687). The DNA template for PCR was first denatured for five minutes at 94°C and then cycled 35 times using the following cycling parameters: 94°C for 30 seconds, 56°C for 30 seconds, 72°C for 60 seconds. Finally, a five minute extension cycle at 72°C was performed. After the second amplification, the PCR products were purified using Qiagen PCR purification kit (Qiagen). The nucleotide sequence of the purified amplicons was determined using the sequitherm excel DNA sequencing kit, (Epicenter Technologies, Madison, WI 53713). Nesting primers end labelled with radioactive ³²P (Amersham Pharmacia Biotech Inc, NJ) were used for sequence analysis. The primers used for sequence determination were as follows. TATAA element of the promoter region: antisense, 5′ GGACACCACT GGGCCCAGCACA 3′ (+59 to +81); exon 1: sense, 5′ AGGAGCAAAGGCGCCATGGCT 3′ (−15 to +6), sense, 5′ GCGTGTCATCAAAACATACAA 3′ (+321 to +341), antisense, 5′ GCTTGCTCAGCATATATCTGG 3′ (intronic primer, 74 nt downstream of 3′ end of exon 1); exon 2: sense, 5′ GTATGTAGTCATCAAAGAATATG 3′ (intronic primer 63 nt upstream of 5′ end of exon 2); exon 3: sense, 5′ TAGTTAGTATAGCAGA 3′ (intronic primer 59 nt upstream of 5′ end of exon 3); exon 4: sense, 5′ GTCCAGCTGTGAAACTCAGA 3′ (intronic primer, 39 nt upstream of 5′ end of exon 4); exon 5: sense, 5′ GAGGATTGTTCATACCACAGG 3′ (intronic primer 36 nt upstream of 5′ end of exon 5); antisense, 5′ TGAATT TAACACTGATTCTGTT 3′ (+1666 to +1687). Nucleotide sequences were determined by the dideoxy chain termination method31 using a cycle sequencing procedure as follows: denaturation, 94°C for 60 seconds, followed by 20 cycles of 94°C for 30 seconds, 56°C for 30 seconds, 72°C for 60 seconds, and 10 cycles of 94°C for 30 seconds, and 72°C for 60 seconds. Sequence abnormality in each sample was confirmed by repeat analysis of an amplicon generated by a second independent PCR.

To determine the consequence of the two newly observed point mutations that predicted single amino acid substitutions (the 1490T>A and 1452G>A mutations, found in families B and C, respectively) on the catalytic activity of UGT1A1, site directed mutagenesis was performed on a plasmid, pSVK3, which contains the entire UGT1A1 coding region.17 The mutagenised UGT1A1 cDNA was excised by partial digestion with EcoRI and complete digestion with XhoI and subcloned into the pcDNA3.1/Zeo(+) (Invitrogen) expression vector and the nucleotide sequence was determined to confirm the introduction of the point mutation. COS-7 cells, grown in 100 mm dishes and at 50-60% confluency, were transfected with the pcDNA3.1/Zeo(+) containing normal or mutagenised UGT1A1 sequences, using DEAE dextran (Amersham Pharmacia Biotech) as previously described.17 Total protein in the cell lysates was quantitated32 and an equal amount of protein from lysates of cells transfected with the different plasmids was subjected to western blot analysis using a monoclonal antibody against UGT1A1 (mab WP1). The immunoreactive bands were detected using a chemiluminescent substrate (Pierce, Rockford, IL).

Expressed UGT1A1 in each cell lysate was quantitated by a sandwich ELISA as previously described.17 Briefly, ELISA plates were coated overnight with WP1 (a monoclonal antibody against UGT1A isoforms), blocked with a solution containing 3% bovine serum albumin and 5% fetal calf serum, and then overlaid with lysates of the transfected cells. A rabbit antibody raised against a synthetic peptide corresponding to a unique region of UGT1A1 (Pab136) was applied. The detection system consisted of a goat anti-rabbit IgG, conjugated with horseradish peroxidase. Based on ELISA quantitation, equal amounts of UGT1A1 from the various cell lysates were assayed for bilirubin-UGT activity in the presence of 80 μmol/l bilirubin and 4.4 mmol/l UDP-glucuronic acid as previously described.17 At the end of incubation for 60 to 120 minutes, the reaction was stopped by adding 0.4 mol/l HCl-glycine buffer (pH 1.8) and pigments were extracted in chloroform/ethanol (1:1 v/v).17 The solvents were evaporated under a stream of nitrogen, the pigments were dissolved in dimethyl sulphoxide and methanol, and analysed by reverse phase high pressure liquid chromatography using a Waters C₁₈μΒondapak column (Waters, Milford, MA) as previously described.33 Formation of bilirubin glucuronides was calculated from the electronically integrated areas under the bilirubin glucuronide peaks.

Brief clinical histories of the four families follow. The genotypes of the TATAA element and coding region ofUGT1A1 are listed in table1.

Family A. Eighteen month old twin girls were brought to the paediatrician for evaluation of jaundice. Both girls had severe neonatal hyperbilirubinaemia, with peak serum bilirubin levels of 456.5 μmol/l and 410.4 μmol/l (>90% “indirect” by van den Bergh reaction) on the fifth postnatal day. Initially the babies were Coomb's test positive and the exaggerated neonatal hyperbilirubinaemia had been attributed to ABO incompatibility. However, the jaundice persisted after their blood samples became Coomb's non-reactive, and a diagnosis of Crigler-Najjar syndrome was suspected after excluding other causes of hyperbilirubinaemia. The twins were placed on phenobarbital and phototherapy. This resulted in a reduction of serum bilirubin concentrations to the current level of 136-171 μmol/l. Unfortunately, the infants had already sustained brain damage, presumably from the high neonatal bilirubin levels. At the age of 18 months, neurological examination of the twins showed spastic cerebral palsy, chorioathetoid movements, and signs of retarded mental development. The affected infants were compound heterozygotes for the Gilbert-type TATAA element and the deletion of the “A” at nt 1223. The frameshift caused by this deletion introduces a premature stop codon at codon 497, predicting the truncation of the protein by 122 amino acids and complete catalytic inactivation of UGT1A1. The father was heterozygous for the structural region mutation but had a normal TATAA sequence. The mother carried a Gilbert-type TATAA element on one allele, but the entire coding region had normal nucleic acid sequence.

Family B. The proband, a male, had persistent scleral icterus, first brought to the attention of the physician at the age of 10 years. The peak serum bilirubin level was 205 μmol/l (>90% “indirect” reacting). An 18 year old sister of the proband was also noted to have milder hyperbilirubinaemia (85.5 μmol/l, >85% “indirect” reacting). The proband and his sister were compound heterozygotes for the Gilbert-type TATAA element and a point mutation 1490T>A in exon 5. The father had a normal TATAA sequence but was heterozygous for the point mutation 1490T>A in exon 5. The mother was heterozygous for the Gilbert-type TATAA element and had normal coding region sequence. The premature termination codon introduced by this mutation is predicted to truncate the carboxy-terminal end of the protein by 37 amino acids. This was confirmed by immunoblot analysis of lysates of COS cells transfected with the mutagenised expression plasmid, pcDNA3.1/zeo(+)- hUGT1A1 (fig1). The expressed truncated UGT1A1 was catalytically inactive.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Western blot analysis. The mutations detected in two of the four families, 1490T>A (family B) and 1452G>A (family C), were introduced in the plasmid pSVK3-UGT1A1 (see text). The mutated and normal UGT1A1 cDNAs were subcloned into pcDNA3.1/Zeo(+) (see text) and transfected into COS cells. The lysate of the cells were checked for the synthesis of UGT1A1 by western blotting as described in the text. Lanes 1, 3, and 4 represent normal UGT1A1 (53 kDa). Lanes 2 and 5 are the mutated forms of the protein synthesised from UGT1A1 cDNAs carrying the mutations 1452G>A and 1490T>A, respectively. In both cases, 1452G>A and 1490T>A, the protein is truncated at its C-terminal end by 52 and 37 amino acids respectively. These mutant forms of the protein migrated faster than their normal counterpart corresponding to their respective truncations.

Family C. The proband was a female infant with persistent hyperbilirubinaemia since birth with a peak level of 376.2 μmol/l (predominantly “indirect” reacting). Phenobarbital treatment was not effective in reducing the serum bilirubin levels and the patient is currently being treated with phototherapy. A diagnosis of CN1 was made. The parents of the infant are first cousins and the father has normal bilirubin levels. The mother had a history of prolonged neonatal hyperbilirubinaemia. Her recent serum bilirubin level was 51.3 μmol/l. Several other relatives of the proband have noticed intermittent yellow discolouration of the sclera, but these samples were not available for genetic analysis. The proband was homozygous for a 1452G>A mutation in exon 5, which predicts the truncation of UGT1A1 at Trp484. As for family B, this was confirmed by Western blot analysis of the expressed mutagenised UGT1A1 (fig 1). The expressed truncated UGT1A1 was catalytically inactive. The father, who was heterozygous for the structural mutation, but had normal UGT1A1 TATAA elements, had normal serum bilirubin levels. In contrast, the mother, who was also heterozygous for the 1452G>A mutation, but carried a Gilbert-type TATAA element on the structurally normal allele, was hyperbilirubinaemic (51 μmol/l).

Family D. Two sisters and their father were studied from this family. One of the sisters had been included in the original study in 1962 describing CN2 as an entity distinct from CN1.2 This patient has had predominantly unconjugated hyperbilirubinaemia since birth, averaging 282 μmol/l. Serum bilirubin concentrations were reduced by 80% after administration of phenobarbital. The sister of this case, who was also studied, had a serum unconjugated bilirubin level of 203.5 μmol/l. The father was also hyperbilirubinaemic with serum bilirubin levels of 60 μmol/l. Both sisters were homozygous for the Gilbert-type promoter and also for a 524T>A mutation that predicted the substitution of leucine by glutamine at amino acid residue 175. The father was homozygous for the Gilbert-type promoter and heterozygous for the 524T>A mutation. We had previously determined the effect of the 524T>A mutation (L175Q, family D), which reduced the UGT1A1 activity towards bilirubin to approximately 38% of normal.17

The findings in the four families described here are examples of the effect of coexistence of a structural mutation ofUGT1A1 with a Gilbert-type promoter abnormality. In families A and B, the fathers of the probands had a normal TATAA element in the promoter on both alleles ofUGT1A1 and were heterozygous for a structural mutation that predicted (in family A) or was shown (in family B) to truncate and completely inactivate UGT1A1. Consistent with the recessive inheritance of CN1, these heterozygous carriers did not have hyperbilirubinaemia. The mothers in both these families were heterozygous for the Gilbert-type promoter, but theUGT1A1 coding regions were structurally normal. In these women also, serum bilirubin levels were normal, which is consistent with recessive inheritance of Gilbert's syndrome. In contrast, the twin infants in family A and the two children in family B who inherited the Gilbert-type promoter abnormality from their mothers and the structural lesions from their fathers were hyperbilirubinaemic. The presence of the Gilbert-type promoter on the structurally normalUGT1A1 allele resulted in hyperbilirubinaemia in the children. In family C, the mother of the affected child with CN1 was heterozygous for the 1452G>A mutation (which was shown to truncate and inactivate UGT1A1) and carried a Gilbert-type TATAA element on the structurally normal allele. Since heterozygosity for the variant promoter or the CN1 type mutation does not cause hyperbilirubinaemia by itself, the increased serum bilirubin levels in the mother again illustrates the effect of interaction of the Gilbert-type promoter and coding region lesions.

In family D, the two sisters studied were homozygous for both the Gilbert-type TATAA element and the structural region abnormality (L175Q), which reduces the UGT1A1 catalytic activity to 38%.17 This level of residual UGT1A1 activity is similar to that seen in subjects with Gilbert syndrome and, therefore, does not explain the relatively high level of hyperbilirubinaemia seen in these sisters. Because the observed bilirubin levels cannot be accounted for by either the structural mutation or the Gilbert-type promoter alone, these cases show that the reduced expression of the mutatedUGT1A1 owing to the presence of a Gilbert-type promoter results in a more severe phenotype, where the expressed enzyme has some residual catalytic activity. In the father, the Gilbert-type promoter on both alleles and a heterozygous 524T>A mutation caused mild hyperbilirubinaemia (serum bilirubin 60 μmol/l). Such mild hyperbilirubinaemia would be compatible with the promoter defect alone, but might have been accentuated by the coexistence of the structural mutation.

Promoter reporter studies indicate that the Gilbert-type promoter reduces the expression of UGT1A1 to 20-30% of normal.18 In cases where one allele produces non-functional UGT1A1 and the expression of the only structurally normal allele is reduced because of the presence of the variant promoter, the residual UGT1A1 activity can be calculated to be only 10-15% of normal. The resulting hyperbilirubinaemia is expected to be more severe than that found in Gilbert's syndrome, but less severe than that seen in Crigler-Najjar syndrome type 1. These considerations and our observed results differ from the conclusion of Ciottiet al 29 that the presence of a Gilbert-type promoter on one allele and a CN1 type mutation on the other allele leads to a CN1 phenotype.

Depending on the nature and location of a mutation, amino acid substitutions can lead to a wide range of serum bilirubin concentrations. At one end of the spectrum, there may be only a small reduction of the catalytic activity of UGT1A1, resulting in unconjugated hyperbilirubinaemia which is mild enough to be compatible with the diagnosis of Gilbert syndrome. At the other end are those that cause the most severe forms of CN2 and CN1. In any case, the coexistence of a Gilbert-type promoter, either on the allele bearing the CN-type mutation or on the structurally normal allele (in the case of a heterozygous mutation), can further aggravate the UGT1A1 deficiency. It should be noted, however, that serum bilirubin levels depend not only on the residual UGT1A1 activity, but also on other factors, such as the rate of bilirubin production. This may explain the strong sex difference in the phenotypic expression of Gilbert's syndrome. The majority of females who are homozygous for the Gilbert-type TATAA element do not have clinical hyperbilirubinaemia, probably because of a lower bilirubin load.23 Consistent with this, the mother in family C had a relatively mild hyperbilirubinaemia (51.3 μmol/l), despite the presence of an inactivating genetic lesion on one allele and a Gilbert-type promoter on the other. The gender effect is also apparent from the data on family B. Although the proband and his sister had identical compound heterozygosity, the sister had lower serum bilirubin levels (table 1). In addition to gender, other known or unknown variables, superimposed on the UGT1A1 genotype, can affect serum bilirubin concentrations. For example, reduced UGT1A1 activity, in combination with a high bilirubin load, can result in higher levels of hyperbilirubinaemia. In the twins in family A, the increased bilirubin load because of ABO incompatibility and the resulting haemolysis led to marked exaggeration of neonatal hyperbilirubinaemia (∼428 μmol/l). After cessation of haemolysis and initiation of phenobarbital therapy serum bilirubin levels stabilised at 136-171 μmol/l. Unfortunately, however, permanent brain damage had already occurred in the affected infants. This case exemplifies the importance of compounding of heterozygosity for a structural mutation with the Gilbert-type TATAA element in the promoter of the structurally normal allele. Because of the very high incidence of Gilbert's syndrome in the general population, the possibility of compound heterozygosity for a structural mutation and the Gilbert-type of TATAA element in the promoter should be considered even if only one of the parents has a family history of inherited jaundice.

Based on a gene frequency of 0.3 for the Gilbert-type promoter, as found in several studies in Europe and the USA,18 20-22 9% of the general population would be homozygous and 42% would be heterozygous for the variant promoter. Thus, approximately 51% of the general population carry a Gilbert-type promoter on at least one allele. Therefore, heterozygous carriers of a structural mutation of UGT1A1 (CN1 or CN2 type) have a relatively high probability of carrying a Gilbert-type promoter on the normal allele. This explains the long standing observation that members of families of patients with CN often exhibit intermediate grades of hyperbilirubinaemia. Although the exact incidence of CN is not known, it appears to be less than one in a million, indicating that the gene frequency is less than 1:1000. Based on the gene frequency for the Gilbert-type promoter and that of the structural mutations, compound heterozygosity for a Gilbert-type promoter and CN-type genetic lesion can be calculated to be over 300 times more common than is homozygosity for a structural region mutation. Consistent with this, in our analysis of over 100 cases of intermediate levels of unconjugated hyperbilirubinaemia (persistently between 51 μmol/l and 308 μmol/l) in whom the clinical history was clearly known, we identified the eight cases reported here, in whom an interaction between the Gilbert-type promoter and the structural region mutation was thought to be the cause of jaundice. In only one other case (not included in this paper) was homozygosity for a structural mutation, in the absence of a promoter mutation, the basis of the intermediate level hyperbilirubinaemia. These findings are different from those reported in a Japanese population by Yamamotoet al.30 In that report, homozygosity or compound heterozygosity for structural region mutations was the cause of jaundice in six cases and in only one case hyperbilirubinaemia was caused by the interaction of a Gilbert-type promoter and a structural region mutation. One possible explanation for the difference in results is that the incidence of the longer TATAA element (A(TA)7TAA) is much lower (gene frequency ∼0.11) in Japan than in the white or black populations (gene frequency ∼0.30).24-26 Furthermore, structural mutations causing mild hyperbilirubinaemia appear to be much more common in Japan.25 34 35 In the American population, however, coexistence of a Gilbert-type promoter and a mutation in the structural region of the UGT1A1 gene appears to be a much more common cause of intermediate grades of hyperbilirubinaemia than homozygosity for a mutation in the coding sequences of the gene. The routine use of intense phototherapy has permitted an increasing number of patients with CN1 or CN2 to survive to ages where pregnancy becomes an option. These patients ask for genetic counselling regarding the probability of inherited jaundice in their offspring and the risk of kernicterus in the newborn period. Our results indicate that both the structural mutation and the promoter genotype of the parents should be taken into account in providing the genetic counselling.