Progressive familial intrahepatic cholestasis (PFIC) is an important group of chronic intrahepatic cholestatic disorders in childhood. PFIC can be divided clinically into 2 phenotypes: those with normal or low serum γ-glutamyl transpeptidase (GGT) activity and those with high GGT. The former includes 2 subtypes, called PFIC1 and PFIC2, caused by ATP8B1 and ABCB11 mutations, respectively. ATP8B1 and ABCB11 mutations are important causes of PFIC and low GGT in Chinese patients, with characteristic mutation spectra (1,2). The latter phenotype, PFIC3, includes disease caused by ABCB4 mutations. PFIC3 can be distinguished from PFIC1 and PFIC2, not only by high GGT but also by histopathologic findings in liver, with ductular proliferation despite patency of intrahepatic and extrahepatic bile ducts; such proliferation is not a feature in PFIC1 or PFIC2 (3–5).
ABCB4, at chromosome 7q21.1, encodes multidrug resistance protein 3 (MDR3) (6,7). MDR3 is almost exclusively expressed in the canalicular membrane of hepatocytes (8). It functions as a phosphatidylcholine floppase, transferring this lipid from the inner to the outer hemileaflet of the lipid bilayer of the canalicular membrane (9). Mutations in ABCB4 lead to functional or absolute MDR3 deficiency, resulting in liver injury; when severe, chronic cholestasis progresses to cirrhosis and liver failure before adulthood (10). ABCB4 mutations have been reported mainly in Europeans (4,11–13) and north Africans (3,14,15). In Asia, few cases have been reported (15–17). We sought to investigate the characteristics of ABCB4 mutations in mainland Chinese children with high GGT chronic intrahepatic cholestasis, and to explore aspects of clinical diagnosis in PFIC3 by comparing clinical features and responses to therapy in patients with and without ABCB4 mutation.
Between March 2004 and October 2009, all of the children referred to the pediatric liver center of Children's Hospital of Fudan University with chronic intrahepatic cholestasis and elevated GGT of unknown cause were considered for study. Included were children with clinical and/or biochemical cholestasis lasting >3 months and with serum GGT activity ≥2 × the upper limit of normal (normal <50 U/L). Excluded were children with the following:
1. Extrahepatic cholestasis or sclerosing cholangitis demonstrated by hepatobiliary imaging studies (ultrasound scan and hepatobiliary iminodiacetic acid cholescintigraphy in every case, and laparotomic/laparoscopic cholangiography or magnetic resonance cholangiopancreatography in selected cases)
2. Serologic or medical-history evidence of congenital infections, parenteral alimentation, drug-related cholestasis, or cholestasis secondary to sepsis or ischemic injury
3. Evidence of metabolic diseases, such as abnormal plasma amino acid or fatty acid metabolites profiles or persistently elevated plasma lactate levels
4. Obvious extrahepatic abnormalities, such as congenital cardiopathy, facial dysmorphism, vertebral defects, ocular abnormalities, and/or renal abnormalities
5. Fibrocystic liver disease (eg, Caroli disease) diagnosed by clinical or other findings (sonography, liver biopsy)
6. Hepatosplenomegaly without other evidence of portal hypertension, suspicion of storage disorders (eg, Niemann-Pick disease, Gaucher disease) or for hemophagocytic lymphohistiocytosis
7. Coagulopathy (international normalized ratio >1.5) that could not be fully corrected after vitamin K intravenous injection at presentation
8. Parents unwilling to take part in the study
Study procedures were carried out in agreement with the 1975 Declaration of Helsinki and approved by the ethics committee of the Children's Hospital of Fudan University. In total, 13 patients from 8 different provinces or municipalities of China were enrolled. Informed consent was obtained from all of the subjects included and/or their parents.
An additional 50 population-matched healthy children undergoing annual physical examination in the setting of hepatitis B vaccination served as controls for the screening of missense mutations, undertaken to exclude the mutations as polymorphisms in the mainland Chinese population.
Genomic DNA of peripheral blood leukocytes was extracted routinely using the Tiangen Blood Genomic DNA Isolation Kit according to the manufacturer's instructions (Tiangen Biotech, Shanghai, China). All of the 27 coding exons of ABCB4 together with at least 100 bp of the adjacent intronic sequence were amplified by polymerase chain reaction (PCR) and directly sequenced in all 13 patients. A list of primers is available upon request. Purified PCR products were detected by laser-induced fluorescence on an ABI Prism 3730 Genetic Analyzer (Applied Biosystems, Foster City, CA). Sequence analysis was performed using BIOEDIT software (North Carolina State University, Raleigh, NC), with results double-checked by 2 of the investigators. Sequence variations known as single-nucleotide polymorphisms (SNPs) were excluded by basic local alignment search tool analysis (http://blast.ncbi.nlm.nih.gov/blast.cgi). Genomic sequences were obtained at the National Center for Biotechnology Information with RefSeq NG_007118.1 as ABCB4 reference. Possible mutations were confirmed by direct sequencing from both ends of a second independent PCR fragment and by direct sequencing in parents.
Homology and Structural Predictions
To estimate the significance of nonclassical splicing mutations, Splice Site Prediction by Neural Network (http://fruitfly.org/seq_tools/splice.html) and MaxEntScan (http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html) software were used. Homology between MDR3 and nonhuman MDR3 orthologues was surveyed using ClustalX software (http://www.ebi.ac.uk). Polyphen (polymorphism phenotyping) software (http://genetics.bwh.harvard.edu/pph) was used to predict the possible impact of amino acid residue substitutions on the structure and function of MDR3 proteins. Polyphen calculates position-specific independent counts (PSIC) scores for 2 amino acid variants in the polymorphic position. PSIC score differences <1.5 denote a benign variant, differences between 1.5 and 2 denote a possibly damaging variant, and differences >2 denote a probably damaging variant (18). Effects of missense mutations on mRNA splicing were analyzed by ESE Finder software (http://rulai.cshl.edu/cgi-bin/tools/ESE3/esefinder.cgi?process=home).
Archival paraffin-embedded liver biopsy specimens were available for immunostaining in 6 of 13 patients. Sections at 4 μm, dried overnight at 37°C on poly-L-lysine-coated slides, were immunostained for MDR3 and, as a control for fixation and processing, for a homologous canalicular transporter, bile salt export pump (BSEP), using a Leica Bond-Max automated immunostainer (Leica Microsystems, Milton Keynes, UK) with mouse monoclonal anti-MDR3 (P3II-26/ALX-801-028; Enzo Life Sciences, Exeter, UK) and rabbit polyclonal anti-BSEP (HPA019035; Sigma-Aldrich, Gillingham, UK). Canalicular marking was evaluated on microscopy by a histopathologist ignorant of both clinical status and results of genetic analysis. Normal liver served as a further control.
Clinical Data and Statistical Analysis
Clinical data for all of the patients were collected by reviewing medical records, including family history, age of onset, presenting symptoms, disease evolution, liver biopsy findings, biochemical parameters at presentation to our hospital and ursodeoxycholic acid (UDCA) response. Patients were divided into 2 groups (with or without ABCB4 mutations). Differences in biochemical parameters between the groups were statistically analyzed. A 2-tailed P value of <0.05 was considered significant.
Characterization of ABCB4 Mutations
Among 13 patients in whom ABCB4 was analyzed, 3 patients harbored ABCB4 mutations and were diagnosed as having PFIC3. No mutation was found in the entire coding sequence, with flanking areas, of ABCB4 in the remaining 10 patients. Nine SNPs detected in the 13 patients had been reported before. No novel SNP was found.
The 3 patients with PFIC3 had 6 distinct ABCB4 mutations, none previously associated with PFIC3. Apart from c.139C>T (p.R47X) (19), they were novel; they included c.344+2_+3insT, c.1376A>G (p.D459G), c.1745G>A (p.R582Q), c.2077_2078delC (p.P693HfsX698), and c.3825_3826delA (p.M1276WfsX1308). Each patient had 2 distinct mutations documented as on different copies of chromosome 22 by segregation analysis, with parents of a particular child each carrying 1 of the 2 mutations found in that child (Table 1).
Predicted Effects of Mutations on MDR3 Function
Nonsense-mediated decay probably rapidly degrades mRNA containing c.139C>T. Translation would generate a truncated protein (p.R47X) including only the first intracellular domain. The stop codon at position 693 caused by c.2077_2078delC would result in a protein truncated after the first 6 transmembrane domains. The normal stop codon would disappear as a result of c.3825_3826delA, with prolongation of MDR3 by an additional 22 amino acid residues (terminated by another stop codon at position 1308). MaxEntScan algorithms predicted splicing mutation c.344+2_+3insT substantially to decrease splice-site scores (maximum entropy model, from 6.43 to −1.66; maximum dependence decomposition model, from 10.08 to 4.38; first-order Markov model, from 5.09 to −0.25; weight matrix model, from 7.20 to 1.41) and splice site prediction by neural network algorithms predicted this mutation to abolish a cryptic donor splice site.
Amino acid residues in the first highly conserved nucleotide-binding domains (NBDs) (residues 404–630) are affected by c.1376A>G (p.D459G) and c.1745G>A (p.R582Q). Nucleotide-binding domains are responsible for binding and hydrolysis of adenosine triphosphate. Cross-species amino acid residue analysis (chimpanzee, ox, rabbit, rat, mouse, fish) showed that both canonical residues are highly conserved. Secondary-structure predictions (Polyphen) gave PSICs of 1.34 and 1.876 for the canonical residues D459 and R582, respectively, and of −0.648 and −0.711 for the variant residues 459G and 582Q, respectively. The absolute differences between the 2 profile score sets are 1.988 and 2.468, respectively, indicating that c.1376A>G (p.D459G) is likely to affect protein function and that c.1745G>A (p.R582Q) is able to affect protein function. ESE Finder predicted c.1376A>G (p.D459G) to decrease the splicing-factor score at a SF (splicing factor)2/alternative splicing factor (ASF)–binding site (from 3.960444 to 3.233662) and to abolish a SC35-binding site; it predicted c.1745G>A (p.R582Q) to enhance the splicing-factor score at a SF2/ASF-binding site (from 3.317899 to 3.765121), to abolish a SF2/ASF-binding site, and to generate a new SRp40-binding site. The results indicate that apart from amino acid substitution, these 2 missense mutations can also affect mRNA splicing. Neither of these missense mutations was detected in the 50 control infants screened (100 control alleles).
In normal liver and in the liver of patients without ABCB4 mutations (3, 5, 6, 8; Table 2), canalicular membranes of hepatocytes marked well for both MDR3 (Fig. 1A and C) and BSEP (Fig. 1B and D). MDR3 staining was completely absent with compound heterozygous mutations c.139C>T (p.R47X) and c.1745G>A (p.R582Q) (case 13, Fig. 1E). Faint MDR3 staining was found with compound heterozygous mutations c.1376A>G (p.D459G) and M1276WfsX1308 (case 11, Fig. 1G). In the 2 patients with ABCB4 mutations, the canalicular transporter BSEP stained normally (Fig. 1F and 1H).
Table 2 summarizes the clinical presentations and outcomes of the 13 patients. Only the parents of case 1 are consanguineous. Cases 4 and 5 were siblings; the other 11 cases are from 11 unrelated families. The mother of case 10 had intrahepatic cholestasis of pregnancy (ICP). Case 11 had an older sister with similar liver disease who died of liver failure at age 6.5 years. Every member of the family of case 12 had hepatobiliary disease: her older sister died of liver failure at age 6 years, their father had symptomatic cholecystolithiasis and gallbladder sludge of early onset (younger than 40 years) and performed cholecystectomy, and ICP was experienced in their mother in the 2 pregnancies.
Intracranial hemorrhage ascribed to delayed vitamin K deficiency occurred in case 11 at age 35 days. Whether cholestasis was then present is unclear. The first noticed liver-related symptoms in case 11 were jaundice, pruritus, and hepatomegaly at age 4.5 years. Portal hypertension, with esophageal varices, was found at age 5 years, with a single episode of gastrointestinal hemorrhage 6 months thereafter. Her weight and height both were at 0 SD (standard deviation) against the standard mainland Chinese growth curve. Her older sister, who experienced a similar disorder, had intracranial hemorrhage ascribed to delayed vitamin K deficiency at age 53 days, with jaundice, pruritus, and hepatosplenomegaly identified at age 6 years. Sudden deterioration 3 months later led to the sister's death in liver failure at age 6.5 years.
In case 12, jaundice was first noticed when she was 1 month old. Her abdomen enlarged progressively thereafter. At age 12 months, she presented with growth retardation, pruritus, and hepatosplenomegaly. At age 16 months, she experienced intracranial hemorrhage ascribed to vitamin K deficiency. From age 17 months, pruritus became more obvious and epistaxis occurred several times. At age 27 months, she was referred to our hospital with jaundice and pruritus. Physical examination revealed hepatosplenomegaly and dilated abdominal wall veins on the abdominal wall. Her weight and height both were at −2 SD against the standard mainland Chinese growth curve. Her older sister, who had a similar disorder, died in liver failure 4 months after admission to our hospital at age 6 years with hepatosplenomegaly, ascites, coagulopathy, and esophageal variceal bleeding.
Case 13 presented with jaundice ages 3 months. Other manifestations included chronic diarrhea and low-grade fever. At age 2 years, pruritus became obvious. Between age 2 and 3 years, he twice broke bones. From age 4 years, gross abdominal distension, hepatosplenomegaly, and growth retardation were evident. At age 9 years, he was referred to our hospital. His firm liver extended 9 cm below the right costal margin and his spleen tip extended 6 cm below the left costal margin. His weight was at −3 SD and his height was at −2.5 SD against the standard mainland Chinese growth curve.
Biochemical parameters of the 13 patients on initial evaluation at our hospital are summarized in Table 3. Total bile acid (TBA) concentrations were significantly higher in patients with ABCB4 mutations than in patients without ABCB4 mutations (352.5 ± 97.0 vs 55.9 ± 50.4 μmol/L, P = 7.32E-05). Other biochemical parameters did not differ statistically between the 2 groups.
Response to Oral UDCA Administration
All of the patients received UDCA orally at the conventional dosage of 15 to 20 mg · kg−1 · day−1 in divided doses given with meals. In the 10 patients without ABCB4 mutations, 5 patients (4–6,8,10) responded well to UDCA therapy: the clinical manifestations of liver disease disappeared gradually and all liver function test results returned to within normal ranges. Case 10 received UDCA treatment for only 3 months, but UDCA administration in the other 4 patients continued for >1 year. Two patients (2,7) responded partially, with milder jaundice and decreased serum alanine aminotransferase, aspartate aminotransferase, GGT, and TBA values that, even after long-term UDCA administration, were not within normal range. Two patients (3,9) were nonresponders; hepatic synthetic function continued to worsen, with death in liver failure after 6 and 12 months, respectively. Case 1 had taken UDCA intermittently for a short time, then underwent liver transplantation. This precluded evaluation of the effect of UDCA therapy.
Among the 3 patients with ABCB4 mutations, responses to UDCA treatment were similar. Pruritus improved significantly, then disappeared, with reduction in scratching, reactive keratosis, and parentally perceived “dryness” of skin. Growth also improved. At last follow-up, weights were at 1 SD in case 11, −1 SD in case 12, and −2 SD in case 13; heights were at 1 SD in case 11, −1 SD in case 12, and −1.5 SD in case 13 against the standard mainland Chinese growth curve. Abdominal-wall venous prominence lessened. Some other clinical and laboratory changes of the 3 patients are listed in Table 4. Generally, after long-term UDCA administration, spleen size decreased and platelet counts increased, with the serum alanine aminotransferase level decreased nearly to normal range, although GGT and TBA values were not improved much.
Published studies on PFIC3 caused by ABCB4 mutations are mostly from Europe (4,11,12). Only 2 patients with PFIC3 diagnosed by genetic testing have been reported from Asia (16,17), although another such patient, of south Asian origins, has been described (15). In the present study, 13 mainland Chinese patients with high-GGT chronic intrahepatic cholestasis were studied. ABCB4 mutations were detected in 3 cases and 5 novel mutations were found. Although they were not genetically tested, the sisters of case 11 and that of case 12, who both had similar clinical manifestations, may also carry the ABCB4 mutations identified in their siblings.
Definitive diagnosis of PFIC3 requires genetic testing. Different studies have identified many novel mutations: 5 of 6 in the present study, 13 of 16 in France (4), and 25 of 29 in Italy (11). These data suggest a wide spectrum of ABCB4 mutations and genetic heterogeneity of the disease, which may create difficulties for clinical genetic testing. The frequency of ABCB4 mutations is not high among patients in whom PFIC3 is suspected clinically: ABCB4 mutations were identified in 3 of 13 patients in this series, 1 of 6 patients in Taiwan (16), 17 of 31 patients in France (4), and 28 of 133 patients in Italy (12), suggesting that the clinical diagnosis of PFIC3 caused by ABCB4 mutations is difficult. To find clues for clinical differential diagnosis, clinical manifestations, biochemical parameters, and response to UDCA therapy were compared between patients with or without ABCB4 mutations.
In contrast, clinical manifestations did not differ substantially between 3 patients with PFIC3 and 10 patients with cholestatic liver disease of unknown etiology. While analyzing differences in biochemical parameters between the groups, we found that TBA levels were significantly higher in patients with ABCB4 mutations than in patients without ABCB4 mutations (P = 7.32E-05). Pruritus in the 3 patients with ABCB4 mutations was serious enough to interrupt the patients’ sleep, although only mild jaundice (total bilirubin <60 μmol/L) and mildly elevated GGT (<200 U/L) (Table 3) existed. In the 10 patients without ABCB4 mutations, pruritus was reported in 4 patients with various jaundice (total bilirubin 10.5–219.0 μmol/L) and markedly elevated GGT (Table 1, cases 2, 3, 7, 9, all >400 U/L). These indicate that if patients have mild jaundice and severe pruritus, with mildly elevated GGT and high TBA, PFIC3 should be strongly suspected. If parents or first-degree relatives have or had symptoms of ABCB4 deficiency such as low phospholipid-associated cholelithiasis or ICP, this would strongly suggest PFIC3.
We could assess MDR3 expression by immunostaining in only 2 of our 3 patients with documented ABCB4 mutations, but in one, MDR3 was wholly lacking and in another, its expression was diminished significantly. These observations encourage us to use liver biopsy and immunostaining studies in evaluating children in whom we suspect ABCB4 disease. The results of such studies may permit the diagnosis of MDR3 deficiency, if not necessarily of ABCB4 mutation.
In the 3 patients with ABCB4 mutations, pruritus disappeared completely and some biochemical parameters improved after UDCA therapy, but TBA levels remained high. Clinical symptoms in all 3 patients improved after UDCA therapy, including disappearance of pruritus, higher quality of life and sleep, and better growth and development, although liver function test results did not completely revert to normal. Examination in all 3 patients suggested cirrhosis before UDCA therapy began. We infer that UDCA therapy can improve ABCB4 disease, or delay its progression, even if cirrhosis has already developed.
One of the limitations of the present study is that liver tissue was not obtained in case 12, who had ABCB4 mutations, so the effects of those mutations on MDR3 expression could not be studied. Nevertheless, considering that the patient was a compound heterozygote for a splicing mutation and for a deletion mutation predicted to impair MDR3 function severely, the diagnosis of PFIC3 caused by ABCB4 mutations is not in serious doubt in this patient. This report is also limited by the relatively small number of cases and the relatively short duration of UDCA treatment. Assessment of long-term prognosis requires a longer observation period, preferably with more study subjects.
In conclusion, ours is the first report to our knowledge to study the significance of ABCB4 mutations in mainland Chinese children with high-GGT chronic cholestatic liver disease. We found that about one-fourth of our cases of this disease were associated with ABCB4 deficiency. UDCA therapy was generally helpful in improving clinical symptoms and liver function in children with this disease, whether or not ABCB4 mutations could be found.
We thank Prof Y.K.L. for the revision and editing of the manuscript. We also thank the patients and their parents.
© 2012 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology,