Mutation in the Sterol 27‐Hydroxylase Gene Associated with Fatal Cholestasis in Infancy

Mutations in the sterol 27-hydroxylase gene (CYP27A1) cause the disease cerebrotendinous xanthomatosis (CTX) characterized by accumulation of cholesterol and cholestanol in tendons, brain and other tissues (1). In the absence of adequate sterol 27-hydroxylase, the normal oxidation of the steroid side chain does not occur and there is a defective bile acid synthesis with excretion of large amounts of 25-hydroxylated C27 bile alcohols in faeces and urine. Because one of the 25-hydroxylated bile alcohols may be converted into cholic acid, total bile acid synthesis is only partially reduced. The disease was first described in 1937 in a patient with dementia, ataxia, cataracts and xanthomas in the Achilles tendons and central nervous system (2). Several hundred patients have been reported in different countries, particularly Japan, Israel and the Netherlands (1). CTX is a slowly progressing disease usually not diagnosed until adult life with the appearance of large xanthomas and dementia. However, a limited number of pediatric patients have been described, often with intractable diarrhea as the presenting symptom (3). Although liver disease is not thought to occur in CTX patients, Clayton et al. (4,5) have recently described a patient with neonatal cholestasis who responded well to treatment with bile acid replacement and whose CTX was not diagnosed until age 7 years. This patient had two siblings with suspected liver disease. Furthermore, in a retrospective analysis, it was reported that prolonged jaundice occurred more often in adult CTX patients than in the general population (5). Thus, the clinical presentation of CYP27A1 mutations is variable. It includes not only the typical neurologic symptoms of adulthood but also gastrointestinal symptoms such as severe diarrhea and neonatal cholestasis (5). We report on an infant with severe cholestatic liver disease in whom the diagnosis of CTX was obtained a few weeks before the patient's death at 4 months of age. PATIENTS AND METHODS Case Presentation The boy was born at term, after an uneventful pregnancy. He was the first child of healthy Bangladeshi parents who were first-degree cousins (the paternal grandmother and the maternal grandfather were siblings). Birth weight was 2.4 kg, suggesting intrauterine growth retardation. Apart from neonatal unconjugated hyperbilirubinemia that was treated with phototherapy, the immediate postnatal period was seen as normal. At 6 weeks of age the boy was referred to the tertiary unit for pediatric hepatology at Huddinge University Hospital because of continued jaundice and conjugated hyperbilirubinemia. On presentation, hepatocellular dysfunction was suspected when a prothrombin level of 40% of normal was noted, which did not improve after the administration of intravenous vitamin K. The serum alkaline phosphatase was markedly elevated, whereas the gamma-glutamyl transferase levels were repeatedly normal throughout the entire course of the disease. Percutaneous liver biopsy showed profound giant cell transformation with an extremely high number of nuclei per hepatocyte, very little lymphocytic infiltration and moderate bile duct proliferation. Cytomegalovirus (CMV) was isolated in urine; CMV-immunoglobulin (Ig)M was detected in the serum; CMV-DNA was detected by polymerase chain reaction (PCR) in white blood cells; no CMV was detected by immunohistochemical analysis in liver tissue. A laparotomy showed patent but very small extrahepatic ducts. At the time, no other cause for cholestasis and ongoing CMV infection could be established despite comprehensive viral and metabolic investigations. Initial treatment consisted of fat-soluble vitamins and ursodeoxycholic acid. Intravenous ganciclovir was given for the CMV infection for 4 weeks, after which CMV-DNA was not detected in white blood cells by PCR. Further examination of the urine collected before the start of ursodeoxycholic acid treatment suggested an inborn error of bile acid synthesis, possibly CTX. When these results were obtained (3 months of age), liver failure had already developed. Treatment with cholic acid was begun. Chenodeoxycholic acid, recommended for patients with CTX (1,3), was avoided because of the potential hepatotoxicity (5). The patient was listed for cadaveric liver transplantation at 3.5 months of age. However, despite intensive treatment, including the use of molecular absorbant recirculating system, hemodialysis and mechanic ventilation, the boy died of multiorgan failure at 4 months of age before a suitable donor organ was available. Unhydrolyzed bile acids and other steroids in urine were analysed by nano-electrospray mass spectrometry using a Micromass Quattro Micro triple quadrupole mass spectrometer (Waters Corp., Milford, MA). Samples dissolved in methanol were electrosprayed from metal-coated borosilicate capillaries (Protana Engineering A/S, Odense, Denmark). The capillary voltage was optimized at approximately −1.6 kV, the cone voltage was −50 V and a counter current of nitrogen (50 L/h) was used to aid solvent evaporation. The source temperature was 70°C. The conditions for sample pretreatment were similar to those previously used in analyses of urine by fast-atom bombardment or electrospray ionization mass spectrometry (6,7). Part of the urine sample was hydrolyzed, derivatized and analyzed by combined gas chromatography-mass spectrometry (8). Assay of Oxysterols in Plasma by Isotope Dilution-Mass Spectrometry Oxysterols in plasma, including 7α-hydroxycholesterol and 27-hydroxycholesterol, were analysed by isotope dilution-mass spectrometry with use of deuterium-labelled internal standards as described previously (9). Mutation Analyses DNA from the infant as well as from both parents was prepared from whole blood using the QIAGEN Blood & Cell Culture DNA Midi kit (Merck Eurolab, Stockholm, Sweden). PCR amplification of all exons, parts of the introns and parts of the promoter was performed according to the protocol given in (Table 1).TABLE 1: Amplification conditions for the PCR analysis of the CYP27AI geneFigure 1 and Table 2 detail extensive primer information. MgCl2-free PCR-buffer, distilled water, dimethyl sulfoxide, Taq-polymerase, free nucleotides and template DNA were added in all samples and the amplification PCR was performed on a MJ Research PTC-225 Peltier Thermal Cycler (Scandinavian Diagnostic Services, Falkenberg, Sweden) using the following program: 94°C for 3 minutes followed by 34 cycles of 94°C for 30 seconds, 59°C for 1 minute, 72°C for 2 minutes, followed by 72°C for 5 minutes and then cooling to 8°C.TABLE 2: Primer information for the PCR analysis of the CYP27A1 geneFIG. 1: Schematic overview of the CYP27A1 gene and primer positions.The PCR product was purified using the QIAGEN QIAquick PCR purification kit for microcentrifuge. The purified DNA fragments were used as template for the sequencing PCR using the Big Dye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA). For sequencing, all amplification primers plus several new primers were used, see Figure 1 and Table 2. The sequencing PCR was performed using the following program: 24 cycles of 95°C 30 seconds, 50°C 15 seconds and 60°C 1 minute, followed by cooling to 4°C. The sequencing PCR products were analysed on a Hitachi 3100 Genetic Analyzer (Applied Biosystems). The study was approved by the local ethics committee at Huddinge University Hospital, and informed consent was obtained from the parents. RESULTS Figure 2 shows an electrospray mass spectrum of a urine sample from the child collected at the age of 8 weeks, before treatment with ursodeoxycholic acid. The most prominent peaks in the m/z range 367.4 to 445.4 correspond to sulphated 3β-hydroxy-Δ5 steroids typical of the urinary steroid pattern in the neonatal period. Peaks attributable to cortisol metabolites were intense (m/z 539.5 and 541.5, representing the glucuronides of tetrahydrocortisone and tetrahydrocortisol, respectively), indicating an increased production of cortisol. Only a small peak corresponding to normal bile acids was seen (at m/z 514, corresponding to trihydroxycholanoyltaurine). This peak of taurocholic acid is frequently one of the major ones in spectra of urine from cholestatic infants (4,7). In the present patient the peak was very small for a cholestatic subject and none of the common peaks of bile acids seen in patients with neonatal cholestasis not attributable to inborn errors of bile acid synthesis were present. However, a small peak of doubly charged dihydroxycholenoyltaurine sulfate at m/z 288.9 (not shown in (Fig. 2) suggested an abnormality of bile acid biosynthesis.FIG. 2: Electrospray mass spectrum of a urine sample from the patient collected at 8 weeks of age. The x-axis shows the mass/charge ratio (m/z) of the ions, the y-axis their relative abundance (%). The insert shows a magnification of the m/z range 580-720 in which ions of glucuronidated bile alcohols and doubly conjugated hydroxycholesterols are seen.A series of peaks of diagnostic importance was seen at m/z 595.5, 611.5 and 627.5. These correspond to glucuronides of cholestanetriols, cholestanetetrols and cholestanepentols known to be excreted in urine of patients with CTX (1,6). The predominant bile alcohol was identified as 5β-cholestane-3α,7α,12α,25-tetrol by combined gas chromatography-mass spectrometry analysis and lesser amounts of 5β-cholestane-3α,7α,12α,24,25-pentol were also identified. If the excretion of these bile alcohols was caused by a lack of sterol 27-hydroxylase, the levels of 27-hydroxycholesterol in the circulation should be low or undetectable (1,4). Combined gas chromatography-mass spectrometry analysis of a plasma sample collected at the age of 8 weeks showed that the level of 27-hydroxycholesterol was 2 ng/mL. In a previously studied population of 7 control infants aged 0 to 1 years, the mean level of 27-hydroxycholesterol was 81 ng/mL with a range of 13 to 158 ng/mL (10). The level of 7α-hydroxycholesterol in the serum of the patient was at the detection limit, approximately 2 ng/mL. Sequence analysis of all nine exons of the genomic DNA originating from the patient demonstrated the presence of a stop codon in exon 7 (G1234T mutation) (Fig. 3). Analysis of genomic DNA from the two parents showed that both were heterozygotes (results not shown).FIG. 3: Mutation analysis of DNA from the patient. The figure shows a G to T mutation in exon 7 at base number 1234 from start of the transcription (arrow). This mutation of the CYP27A1 gene leads to a change of amino acid n:o 408 from glutamate to a stop codon.DISCUSSION Our patient had mutation in the sterol 27-hydroxylase gene that has not been described previously. This mutation would be expected to lead to a truncated and inactive enzyme, explaining the very low levels of 27-hydroxycholesterol and the accumulation of C27 bile alcohols. In our patient this was associated with the development of early and severe cholestasis. Similarities in clinical presentation have been reported between CTX patients with the same genetic mutations (11). However, the patient with neonatal cholestasis described by Clayton et al. (5,11) had the same mutation as a patient with xanthomas and cataract but without cholestasis. Although the mutation found in our patient is unique, this does not explain the very different clinical development in our patient compared with most other patients with CTX (12). Our patient had an ongoing CMV infection, which is a recognized cause of neonatal cholestasis (13-15). It is possible that the combined effects of the viral infection and the metabolic defect of CTX produced a rapidly progressive severe liver disease. It has been shown that such a combination of viral disease and metabolic defect may aggravate the cholestasis in patients with Δ4-3-oxo-steroid 5β-reductase deficiency (16). The electrospray mass spectra did not show significant peaks of such bile acids in the present case. Also, except for a small peak corresponding to taurocholate, none of the most abundant peaks seen in cases of nonspecific cases of neonatal cholestasis were present. This was an indication of a defect in bile acid biosynthesis. Although the occurrence of giant cells is a non-specific, age-related phenomenon in intrahepatic infantile cholestasis, the unusually high number of nuclei per hepatocyte might indicate a primary error of bile acid synthesis (4,17,18). The relative lack of lymphocytic aggregates and the presence of bile duct proliferation rather than paucity of intrahepatic bile ducts in the liver biopsy would argue against CMV infection as the primary cause of the our patient's liver disease (15). The rapidly progressive course despite successful eradication of CMV also suggests that the metabolic defect was primary to the child's liver disease. Based on findings in one infant lacking oxysterol 7α-hydroxylase (CYP7B1), Setchell et al. (19) concluded that the gene coding for the rate-limiting step in degradation of cholesterol into bile acids, cholesterol 7α-hydroxylase (CYP7A1) is not expressed in early infancy. In accordance with this conclusion we found that the plasma level of 7α-hydroxycholesterol, believed to be a marker for the activity of CYP7A1 in the liver (20), was very low in our patient, less than one tenth of the levels in healthy adults. The levels of 7α-hydroxycholesterol in healthy adults are approximately 40 ng/mL with a broad range from 2 to 248 ng/mL reported in 31 healthy adults (9). No information has been published on the levels in healthy infants during the first year of life, but according to unpublished observations in our laboratory the levels are considerably lower than those in adults. In adult patients with CTX there is an induction of cholesterol 7α-hydroxylase as a consequence of the low formation of chenodeoxycholic acid and adult untreated patients with CTX thus have markedly increased levels of 7α-hydroxycholesterol in the circulation (I. Björkhem, unpublished observations). Such an induction was not observed in our patient. The most important alternative pathway to bile acids starts with a 27-hydroxylation of cholesterol and is thus dependent upon the sterol 27-hydroxylase that is deficient in CTX. As a consequence of reduced activity of both CYP7A1 and CYP27A1, neonatal patients with CTX would hypothetically produce very low amounts of normal bile acids. In accordance with this expectation, only traces of a normal C24 bile acid was seen in urine and serum of our patient. The reduced bile acid pool size in the neonatal state (21), particularly in neonates with CTX, may increase the risk for cholestasis. This risk may increase in the presence of additional risk factors for the development of cholestasis (e.g., a viral infection and a secondary deficiency of 3-oxo-Δ4-steroid 5β-dehydrogenase) (22). In view of the low or absent CYP7A1 activity in our patient, the formation of the 7a-hydroxylated bile alcohols excreted in urine may involve the participation of another cytochrome P450 enzyme. The most likely candidate would be CYP7B1, which can use 25-hydroxycholesterol as a substrate. A possible pathway would be initial microsomal 25-hydroxylation of cholesterol followed by 7α-hydroxylation. It could be speculated that this sequence is less efficient than initial 7α-hydroxylation followed by 25-hydroxylation and could constitute a difference between CTX patients who do and do not develop neonatal liver disease. Analysis of hydroxycholesterols in plasma from the two types of patients might reveal such a difference in pathways. The mechanisms by which CYP27A mutations cause cholestasis may also involve other regulatory systems. For example, there is a reduced production of chenodeoxycholic acid in CTX patients. This can lead to a decreased activation of the nuclear farnesoid X-activated receptor, which in turn could result in a reduced expression of the bile salt export pump, causing a decrease in the canalicular bile salt transportation (5,23). A notable difference between the urinary bile alcohol profiles of our patient and previously studied adult CTX patients without liver disease is that cholestanetetrols is predominant in the former while cholestanepentols are the major bile alcohols in the latter (6). Interestingly, in Clayton et al.'s first description of the cholestatic infant (4,5), the urinary bile alcohol profile resembled that of our patient, whereas the pattern changed into a more adult one as the child became older. In the family described by Clayton et al. (4,5) one sibling was stillborn and another died at 13 months having been jaundiced all his life. The third sibling developed neonatal cholestasis and was possibly saved by early introduction of bile acid therapy. His CTX diagnosis was genetically established at the age of 11 years. Our patient was first treated with ursodeoxycholic acid and later also with cholic acid, but it is possible that this treatment was introduced too late to reverse the fatal course. Earlier introduction of bile acid treatment could possibly have altered the clinical course. Because neonatal cholestasis caused by defects in bile acid synthesis may be successfully treated with bile acids (1,5,17,18), it is important to analyze urine from cholestatic patients for the presence of abnormal bile acid or bile alcohol patterns. This is conveniently done by electrospray mass spectrometry. In our case, the combination of severe cholestasis, very low levels of urinary bile acids and normal serum levels of γ-glutamyl transferase further raised the suspicion of an inborn error of bile acid synthesis. The absence of elevated serum γ-glutamyl transferase despite ongoing cholestasis in such patients has been described previously (24). We can only speculate that early bile acid treatment might prevent an infant with CTX from developing xanthomas and neurologic symptoms later in life.