ATP8B1 and ABCB11 analysis in 62 children with normal gamma-glutamyl transferase progressive familial intrahepatic cholestasis (PFIC): Phenotypic differences between PFIC1 and PFIC2 and natural history


  • Anne Davit-Spraul,

    1. Biochemistry Unit, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Monique Fabre,

    1. Pathology Unit, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    2. Faculty of Medicine Paris-Sud 11, University Paris, Paris, France
    Search for more papers by this author
  • Sophie Branchereau,

    1. Pediatric Surgery Unit, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Christiane Baussan,

    1. Biochemistry Unit, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Emmanuel Gonzales,

    1. Faculty of Medicine Paris-Sud 11, University Paris, Paris, France
    2. Pediatric Hepatology Unit and National Reference Centre for Biliary Atresia, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    3. INSERM U757, University Paris 11, Orsay, France
    Search for more papers by this author
  • Bruno Stieger,

    1. Division of Clinical Pharmacology and Toxicology, Department of Medicine, University Hospital, Zurich, Switzerland
    Search for more papers by this author
  • Olivier Bernard,

    1. Faculty of Medicine Paris-Sud 11, University Paris, Paris, France
    2. Pediatric Hepatology Unit and National Reference Centre for Biliary Atresia, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Emmanuel Jacquemin

    Corresponding author
    1. Faculty of Medicine Paris-Sud 11, University Paris, Paris, France
    2. Pediatric Hepatology Unit and National Reference Centre for Biliary Atresia, Hôpital Bicêtre, Assistance Publique, Hôpitaux de Paris, Paris, France
    3. INSERM U757, University Paris 11, Orsay, France
    • Service d'Hépatologie Pédiatrique, Hôpital Bicêtre, 78, rue du Général Leclerc, Le Kremlin-Bicêtre, 94275 cedex, France
    Search for more papers by this author
    • fax: +33 1 45 21 28 16

  • Supported by Assistance Publique, Hôpitaux de Paris and Ministère de la Santé (Paris, France); grants: Innovations Technologiques 2002/2003, Progrès Médical 2004/2005 and DHOS 2005/2006.

  • Potential conflict of interest: Nothing to report. Additional Supporting Information may be found in the online version of this article.


Progressive familial intrahepatic cholestasis (PFIC) types 1 and 2 are characterized by normal serum gamma-glutamyl transferase (GGT) activity and are due to mutations in ATP8B1 (encoding FIC1) and ABCB11 (encoding bile salt export pump [BSEP]), respectively. Our goal was to evaluate the features that may distinguish PFIC1 from PFIC2 and ease their diagnosis. We retrospectively reviewed charts of 62 children with normal-GGT PFIC in whom a search for ATP8B1 and/or ABCB11 mutation, liver BSEP immunostaining, and/or bile analysis were performed. Based on genetic testing, 13 patients were PFIC1 and 39 PFIC2. The PFIC origin remained unknown in 10 cases. PFIC2 patients had a higher tendency to develop neonatal cholestasis. High serum alanine aminotransferase and alphafetoprotein levels, severe lobular lesions with giant hepatocytes, early liver failure, cholelithiasis, hepatocellular carcinoma, very low biliary bile acid concentration, and negative BSEP canalicular staining suggest PFIC2, whereas an absence of these signs and/or presence of extrahepatic manifestations suggest PFIC1. The PFIC1 and PFIC2 phenotypes were not clearly correlated with mutation types, but we found tendencies for a better prognosis and response to ursodeoxycholic acid (UDCA) or biliary diversion (BD) in a few children with missense mutations. Combination of UDCA, BD, and liver transplantation allowed 87% of normal-GGT PFIC patients to be alive at a median age of 10.5 years (1-36), half of them without liver transplantation. Conclusion: PFIC1 and PFIC2 differ clinically, biochemically, and histologically at presentation and/or during the disease course. A small proportion of normal-GGT PFIC is likely not due to ATP8B1 or ABCB11 mutations. (HEPATOLOGY 2010)

Mutations in ATP8B1 or ABCB11 are involved in autosomal recessive hereditary cholestasis with normal serum gamma-glutamyl transferase (GGT) activity.1ATP8B1 encodes familial intrahepatic cholestasis 1 (FIC1) protein, a membrane P-type APase expressed in many tissues that is likely an aminophospholipid flippase.2, 3 Impaired FIC1 function may lead to loss of lipid asymmetry in canalicular membranes of hepatocyte resulting in dysfunction of the bile salt export pump (BSEP), a liver-specific canalicular protein, responsible for biliary bile acid (BA) secretion and encoded by ABCB11.4, 5 Both primary FIC1 and BSEP deficiencies lead to hepatocyte BA overload and are involved in severe as well as milder or benign phenotypes.1-3, 5-12 Patients with a severe phenotype fit in the category of progressive familial intrahepatic cholestasis (PFIC), type 1 in case of FIC1 deficiency, or type 2 in case of BSEP deficiency.2, 3, 5, 7, 8-10 A few reports since the advent of ATP8B1 and ABCB11 genotyping have suggested that clinical differences may exist between the two types of normal-GGT PFIC.1, 7, 8, 9, 13-18 To establish a comprehensive comparison between these two diseases we studied a cohort of 62 normal-GGT PFIC patients in whom ATP8B1 or ABCB11 mutations were searched in an attempt to define their respective clinical presentations, outcomes, and responses to treatment.


BA, bile acid; BD, external biliary diversion; BRIC, benign recurrent intrahepatic cholestasis; BSEP, bile salt export pump; FIC1, familial intrahepatic cholestasis 1; GGT, gamma-glutamyl transferase; HCC, hepatocellular carcinoma; LF, liver failure; LT, liver transplantation; NLT, normal serum liver tests; PFIC, progressive familial intrahepatic cholestasis; sALT, serum alanine aminotransferase; sAFP, serum alphafetoprotein; SNP, single nucleotide polymorphism; UDCA, ursodeoxycholic acid; ULN, upper limit of normal range.

Patients and Methods


We retrospectively reviewed the charts of 62 children (57 families) with unremitting hepatocellular cholestasis, with pruritus, elevated serum BA concentration (>15 μmol/L), and consistently normal serum GGT activity (normal-GGT PFIC), and in whom other causes of cholestasis were excluded, including inborn errors in BA synthesis.13-15, 18-20 Children were referred to the pediatric hepatology unit of Bicêtre Hospital between 1978 and 2007 and had a DNA sample available. Fifty-three (85%) were investigated and followed at this center. Fifteen patients have been reported in part.8, 9, 19, 20 Twenty families were consanguineous. Four siblings in four different families died previously of similar liver disease. Three children (PFIC2 #34-36) had initially a normal-GGT benign recurrent intrahepatic cholestasis (BRIC) phenotype but evolved to a PFIC phenotype.6, 7 Due to their initial phenotype, these three children were not included in the phenotypic comparison analysis. Clinical, biochemical, ultrasonographic, and histological data were recorded. For biochemical variables, the earliest measurement within the first year of life was considered. For analysis of liver histology, data obtained within the first year of life were reported separately from those obtained after age 1 year. Liver sections were analyzed by the same pathologist (M.F.), who was blinded to the results of the genetic study. After 1990, all patients (n = 55) received ursodeoxycholic acid (UDCA) at a daily dose of 600 mg/m2 of body surface.19 Seven patients did not receive UDCA because of early death and/or presentation before 1990. Normalization of serum liver tests (NLT) (transaminases, total bilirubin) defined a positive response to UDCA, no improvement, and partial improvement defined a negative and intermediate response, respectively. In two patients lost to follow-up the response to UDCA was not evaluated. After 1995, external biliary diversion (BD) was proposed in all patients without liver failure (LF) and/or hepatocellular carcinoma (HCC), and with refractory pruritus despite high daily doses of rifampicin (20 mg/kg of body weight) and UDCA (n = 24), and was performed in 15 (nine family refusals).21, 22 BD success was defined as disappearance of pruritus and jaundice, failure as no effect on pruritus nor jaundice, and partial failure as a decrease of pruritus without clearance of jaundice. Liver transplantation (LT) was performed in 25 patients, directly before the UDCA and BD eras, or after failure of UDCA therapy followed or not by BD. Five patients were waiting for LT. LT indications were severe cholestasis, LF, or HCC. Informed written consent approved by the local ethical committee was obtained from each family included in the study.

Genomic DNA Sequence Analysis.

DNA of the 62 patients and their parents was extracted from EDTA blood (QIAampDNA Mini kit, Qiagen, Courtaboeuf, France). Mutation screening was done using polymerase chain reaction (PCR) amplification and DNA sequencing of coding exons 2 to 28 and all splice junctions of ATP8B1 (RefSeq NM_005603.3) and/or ABCB11 (RefSeq NM_003742.2).2, 3, 5-8 Identification and localization of mutations were assessed by sequence comparison using seqscape Software (Applied Biosystems). In the absence of evidence for a disease-causing mutation within the first gene, the second gene was sequenced. Mendelian inheritance was demonstrated by sequencing affected exons in both parents. Missense changes were distinguished from single nucleotide polymorphism (SNP) according to published data and/or public databases (; Combined in silico splice tools were used to test the potential effect of new synonymous or intronic variation on pre-messenger RNA (mRNA) splicing.23 We used the recommendations for mutation nomenclature ( to name ATP8B1 and ABCB11 variations, where A of the ATG translation start codon is +1. In case of new missense mutation identification, 100 control chromosomes were screened to exclude a SNP.

Liver Immunohistochemistry.

Nontumorous liver samples of 35 patients were studied. Anti-BSEP antibody was used as described.7, 24 Anti-MDR3 antibody (Sigma-Aldrich) was used as control.25 Immunohistochemistry was performed after standard liver section analysis and its evaluation was blinded to genotyping result.

Bile Analysis.

Gallbladder bile samples from 24 patients were obtained by percutaneous transhepatic cholecystography or at time of BD or LT. Post-BD, bile was obtained from output into an ostomy bag. Concentration of BAs was measured using standard procedures.20, 25


Comparison of biochemical values between PFIC1 and PFIC2 were performed by the nonparametric Mann-Whitney test. P ≤ 0.05 was considered significant.


Patient characteristics are summarized in Supporting Table 1. ATP8B1 and ABCB11 sequence analyses showed that 13 PFIC patients were type 1 (21%) (PFIC1 families #1-12, Supporting Table 1A) and 39 were type 2 (63%) (PFIC2 families #1-36, Supporting Table 1B). PFIC origin remained unknown (PFIC?) in 10 patients (16%) (PFIC? families #1-9, Supporting Table 1C). Global analysis showed that at least 54 of the 62 patients (87%) are alive at a median age of 10.5 years (range: 1-36).

Mutation Screening and Canalicular Immunostaining

ATP8B1 Mutations (PFIC1).

These were identified in 12 (21%) of the 57 families. Sequence analysis revealed 18 different mutations including eight novel (Supporting Table 2). The 18 mutations included three nonsense mutations, a 7-base duplication leading to truncated protein, a 3-base deletion, a large deletion, and three intronic mutations predicted or known to affect splicing.2 The nine remaining mutations were missense substitutions. In addition, SNP p.R952Q was present on 13% of PFIC alleles.3 The ATP8B1 defect was characterized on both alleles in all PFIC1 families (Table 1). The probands were compound heterozygous or homozygous in six families each. Among patients with missense mutations only, Patient 11 was the only one having a novel mutation on both alleles, which was predicted to be damaging ( Immunostaining studied in five patients showed normal BSEP canalicular staining in four and faint staining in one (Supporting Fig. 1). In all patients MDR3 canalicular staining was normal.

Table 1. ATP8B1 Mutations in PFIC1 Patients
FamilyMutation Allele 1Mutation Allele 2Biliary BA mmol/L (N > 10 mmol/L)BSEP Liver Immunohistochemistry
  • Novel mutations described in this study are shown in bold.

  • *

    Patient harboring the heterozygous ATP8B1 mutation p.R952Q.

  • Patient harboring the heterozygous ABCB11 mutation p.V444A.

  • BSEP ±, faint canalicular staining; BSEP +, normal canalicular staining. BA, bile acids; na, not available.

PFIC1 no. 1p.G446Rp.Y757X3BSEP +
PFIC1 no. 2p.del Ile645-Ile699p.S994RnaBSEP +
PFIC1 no. 3p.G446Rc.782 -1G>Anana
PFIC1 no. 4a*,4b*p.R525Xdel PHE ex 15nana
PFIC1 no. 5p.E665Xp.Tyr1209Leufs28Xnana
PFIC1 no. 6p.del Ile645-Ile699p.del Ile645-Ile699naBSEP ±
PFIC1 no. 7p.G308Vp.G308Vnana
PFIC1 no. 8g.24774-42062delg.24774-42062delnaBSEP +
PFIC1 no. 9p.R600Wp.R600Wnana
PFIC1 no. 10p.P209T + c.625+5 g>tp.P209T + c.625+5 g>t7.9BSEP +
PFIC1 no. 11p.T717Np.T717Nnana
PFIC1 no. 12*p.I661Tp.G702Rnana

ABCB11 Mutations (PFIC2).

These were identified in 36 (63%) of the 57 families. Sequence analysis of ABCB11 revealed 41 different mutations (Supporting Table 3). Twenty-eight mutations were novel. The 41 mutations included six nonsense mutations, seven small insertions or deletions leading to truncated protein, three intronic mutations, and one novel synonymous mutation (p.R1001R) predicted to affect splicing. The 24 remaining mutations were single missense substitutions. In addition to the 41 disease-causing mutations, we found three other variants (p.V444A, p.G319G, p.M677V) considered SNPs.7, 27 The p.V444A mutation was present on 44% of PFIC alleles screened. In patients having missense mutations on both alleles, BSEP immunostaining and/or biliary BA concentration were not available in six, including four harboring a new missense mutation on both alleles. These new missense mutations were predicted to be damaging. ABCB11 defect was characterized on both alleles in 29 cases (Table 2). When mutations were identified on both alleles the probands were compound heterozygous in 17 cases and homozygous for a single mutation in 12. In seven children only one allele was found mutated. However, these patients were considered PFIC2 when the mutation led to a truncated protein, or when a missense mutation was associated with an absence of BSEP canalicular staining and/or low biliary BA concentration. The p.R698H mutation previously reported as a possible rare SNP was identified in three children.11, 12 However, BSEP canalicular staining was negative or focal negative in these three patients, and biliary BA concentration was low in one of them; two were heterozygous for p.R698H and one was heterozygous compound. BSEP canalicular immunostaining performed in 24 patients was negative in 22 and focally negative in two (Supporting Fig. 1). In all patients MDR3 canalicular staining was normal.

Table 2. ABCB11 Mutations in PFIC2 Patients
FamilyMutation Allele 1Mutation Allele 2Biliary BA (N > 10 mmol/L)BSEP Liver Immunohistochemistry
  • Novel mutations described in this study are shown in bold.

  • *

    Patient harboring the heterozygous ABCB11 mutation p.V444A.

  • Patient harboring the homozygous ABCB11 mutation p.V444A.

  • Patient harboring the heterozygous ABCB11 mutation p.M677V.

  • §

    Patient harboring the homozygous ATP8B1 mutation p.R952Q.

  • Patient harboring compound heterozygous status for ATP8B1 mutations p.R952Q and p.E429A.

  • BSEP −, no canalicular staining; BSEP ±, focal positive and negative canalicular staining; BA, bile acids; na, not available; nf, not found.

PFIC2 no. 1p.R1090Xp.R1090X0.02BSEP −
PFIC2 no. 2p.Y354Xp.Y354X0.02BSEP −
PFIC2 no. 3p.Y354Xp.Y354X0.19BSEP −
PFIC2 no. 4p.Y354Xp.Y354Xnana
PFIC2 no. 5ap.Y354Xp.G982R0.10BSEP −
PFIC2 no. 5bp.Y354Xp.G982Rnana
PFIC2 no. 6p.R415Xp.R415Xnana
PFIC2 no. 7p.S226Lp.S226Lnana
PFIC2 no. 8p.G238Vp.G238V0.06BSEP −
PFIC2 no. 9p.E297Gp.E297G0.80BSEP −
PFIC2 no. 10ap.R1128Cp.R1128C0.10BSEP −
PFIC2 no. 10bp.R1128Cp.R1128C0.58BSEP −
PFIC2 no. 11p.R1128Cp.R1128Cnana
PFIC2 no. 12p.T1210Pp.T1210P0.11BSEP −
PFIC2 no. 13c.3213 +5 G>Ap.A1192EfsX50naBSEP −
PFIC2 no. 14ap.I420Tp.I1061VfsX340.02na
PFIC2 no. 14bp.I420Tp.I1061VfsX34nana
PFIC2 no. 15*p.A167Tp.G1058HfsX380.5BSEP −
PFIC2 no. 16*p.R1231Wp.I528Xnana
PFIC2 no. 17p.M62Kp.I112T + p.R698H0.10BSEP −
PFIC2 no. 18*p.E297Gp.H484RfsX50.16BSEP −
PFIC2 no. 19*p.E297Gp.I610GfsX450.23BSEP −
PFIC2 no. 20p.A257Gp.G982Rnana
PFIC2 no. 21*p.I182Kc.3213 +5 G>AnaBSEP −
PFIC2 no. 22p.D549Vc.76 +3 G>Tnana
PFIC2 no. 23§p.M183Tp.G455Enana
PFIC2 no. 24p.R1153Cc.3213 +4 A>G0.13BSEP −
PFIC2 no. 25*p.G982Rp.Q101DfsX80.10BSEP −
PFIC2 no. 26*p.N591S + p.V597Vnf0.39BSEP −
PFIC2 no. 27*p.G982Rp.R1001RnaBSEP −
PFIC2 no. 28p.L232CfsX9nfnaBSEP −
PFIC2 no. 29p.W114Rnf0.50BSEP −
PFIC2 no. 30p.Y1041Xnf0.22BSEP −
PFIC2 no. 31*p.R470Xnfnana
PFIC2 no. 32*p.R698HnfnaBSEP ±
PFIC2 no. 33p.R698HnfnaBSEP ±
PFIC2 no. 34p.M1Vp.R387Hnana
PFIC2 no. 35p.E297Gp.S699Pnana
PFIC2 no. 36p.G1116Fp.R387Hnana

Clinical, Biochemical, and Histological Phenotypes Associated with ATP8B1 or ABCB11 Defect

Comparison Between PFIC1 and PFIC2 Phenotypes.

Clinical features (Table 3).

The first signs of cholestasis usually appear before 3 months of age and tend to appear earlier in PFIC2 than in PFIC1. No clear-cut differences in jaundice evolution pattern were seen between PFIC1 and PFIC2 nor in the development of pruritus, hepatomegaly, or splenomegaly. More than half of the 13 PFIC1 patients presented with extrahepatic symptoms during disease evolution, such as diarrhea and pancreatitis, and a sweat chloride test was elevated in two children.8 LT was performed in six PFIC1 children because of severe cholestasis without signs of LF or liver cancer. Post-LT outcome was complicated in all six children by one or several extrahepatic conditions including severe diarrhea somewhat improved by cholestyramine treatment, exocrine pancreatic insufficiency, deafness, liver steatosis, and lack of catch-up growth in height (Table 3; Supporting Table 1A). PFIC2 patients were notable for the early presence of signs of fat soluble vitamin deficiencies, including rickets or bleeding, and for cholelithiasis that was present in 10 children. Many PFIC2 children progressed early to severe outcome of liver disease, such as LF and/or HCC, which were present at the time of LT in 11 of 19 patients. Signs of LF were observed before 1 year of age in three children and HCC developed as early as 7 months of age in one child (Supporting document, Table 1B).

Table 3. Clinical Comparative Analysis in PFIC1 and PFIC2 Children
 PFIC1 (n = 13)PFIC2 (n = 36)
 n (%)n (%)
  • J, jaundice; DS, discolored stools; H, hepatomegaly; P, pruritus; LT, liver transplantation; SC, severe cholestasis; LF, liver failure; HCC, hepatocellular carcinoma.

  • *

    Two patients had a positive response for 5 and 8.5 years, then jaundice and pruritus reappeared.

  • Four patients are waiting for LT.

Age at first symptoms (median, [range]; month)2 [1-5]2 [1-60]
 First symptoms (J, DS and/or H) ≤1 month2 (15%)16 (44%)
 First symptoms (J, DS, H  and/or P) ≤3 months8 (61%)26 (72%)
 Pruritus13 (100%)36 (100%)
 Hepatomegaly13 (100%)35 (97%)
 Splenomegaly4 (31%)15 (41%)
 Only at presentation04 (11%)
 Recurrent1 (8%)3 (8%)
 Recurrent then permanent3 (23%)9 (25%)
 Permanent9 (69%)17 (47%)
 No jaundice03 (8%)
Signs of vitamin D deficiency (Rickets)01 (3%)
Signs of vitamin K deficiency (Bleeding)03 (8%)
Cholelithiasis010 (28%)
Initial extrahepatic signs  
 Diarrhea8 (61%)0
 Pancreatitis1 (8%)0
 Elevated sweat chlorure test2 (15%)0
UDCA response  
 Positive/mean duration (year)5 (38%)/7.4 y9* (31%)/8.2 y
 Intermediate4 (31%)12 (41%)
 Negative4 (31%)8 (28%)
 Unknown or not tested07
Liver transplantation indication6 (46%)19 (53%)
 SC6 (100%)8 (42%)
 SC + LF06 (31%)
 SC + LF+ HCC05 (26%)
Extrahepatic signs post-LT  
 Diarrhea5 (83%)0
 Liver steatosis3 (50%)0
 No catch-up of stature growth4 (67%)0
 Deafness2 (33%)0
 Pancreatitis2 (33%)0
Laboratory Evaluations (Table 4).

In order to compare PFIC1 and PFIC2, the main biochemical features were recorded at the time of diagnosis in 41 children in whom they were available within the first year of age. Clear-cut differences concerned initial serum alanine aminotransferase (sALT) activities that were higher in children with PFIC2, and serum alphafetoprotein (sAFP) concentrations that were above normal values for age in 12/15 PFIC2 children studied, whereas they were normal in all eight PFIC1 children studied. A combination of sALT and sAFP values obtained simultaneously in 26 patients, within the first year of life, show that all PFIC1 patients presented with sALT ≤5× upper limit of normal range (ULN) and normal sAFP values, whereas 73% of PFIC2 patients presented with sALT >5× ULN and elevated sAFP values. The combined results of sALT >5× ULN and sAFP value above normal for age reflected a negative predictive value of 100% for PFIC1 and a positive predictive value of 92% for PFIC2. Only one patient with unidentified genotype (PFIC? #9) presented with this combination and had a phenotype in favor of PFIC2 (see below). Biliary BA concentrations were dramatically decreased (<1 mmol/L; N > 10) in PFIC2 patients and only mildly decreased (3-7.9 mmol/L) in PFIC1 patients (Tables 1, 2). 4

Table 4. Biochemical Features at Time of Diagnosis Within the First Year of Life in 41 PFIC1 and PFIC2 Patients
  Age at Prothrombin    TotalSerum
  Blood TestAFPTimeFactor VCholesterolALTGGTBilirubinBile Acids
    N > 70%N > 80%N < 5.2N < 40N < 30N < 17N < 15
  • Age at first blood biochemical data (median [range] values).

  • Normal ranges for serum GGT values: <2 months: <230 U/L; 2-4 months: <100 U/L; 4-7 months: <45 U/L; >7 months: <30 U/L. Comparison vs. PFIC1: °not significant; *P < 0.05; **P < 0.005; ***P < 0.0001.

  • n indicates the number of subjects for whom data were available.

  • AFP, alfafetoprotein; AFP value was transformed into multiples of upper limit of normal range for age (ULN); ALT, alanine aminotransferase activity; GGT, gammaglutamyltransferase activity.

PFIC1 (n=12)n1289781212127
PFIC2 (n=29)n291525222229292921
Histological Signs (Table 5).

Liver histology obtained in 30 patients at diagnosis and before age 1 year showed no or mild portal and lobular fibrosis, no necrosis, and no or few giant transformation of hepatocytes in PFIC1 patients, whereas mild to severe portal and lobular fibrosis with bridging fibrosis, cirrhosis, presence of numerous giant hepatocytes, and sometimes necrosis were observed in PFIC2 patients. Liver histology studied after 1 year of age in 31 patients showed major portal and lobular fibrosis in both PFIC1 and -2 patients but lesions were more pronounced in PFIC2 patients, as >50% of them presented with cirrhosis. The more severe pattern in PFIC2 was also illustrated by signs of necrosis, persistence of giant hepatocytes, dysplasia, and the presence of HCC. Both PFIC1 and -2 patients had hepatocellular and/or canalicular cholestasis with various degrees of ductular reaction indicating biliary metaplasia of periportal hepatocytes. 5

Table 5. Liver Histology in PFIC1 and PFIC2 Patients Before and After the Age of 1 Year
PFIC1Portal FibrosisLobular LesionsCholestasis
FibrosisGiant HepatocytesNecrosisHCC DysplasiaLocalizationDuctular Reaction
No.Portal TractSeptalCirrhosis012
  1. Portal fibrosis is reported as, no fibrosis, fibrosis only localized to the portal tract, septal fibrosis without cirrhosis, or cirrhosis. Lobular fibrosis is reported as sinusoidal and central fibrosis and was scored on a scale of 0 to 2, with 0 indicating absence of fibrosis, 1 moderate fibrosis, 2 marked fibrosis. Lobular characteristics: giant hepatocytes: +, numerous, +/−, few, −, absence; HCC, hepatocellular carcinoma. Cholestasis characteristics: h, hepatocellular; c, canalicular; ductular reaction was evaluated using cytokeratin 7 and 19 immunostained sections and recorded as: −, absence; +, presence.

  2. *n indicates the number of subjects for whom data were available.

*n=9, < 1 yr4500270−, 6−,90h, 1−, 3
median age [range]:+/−, 2c, 4+, 6
5.5 mo (2-11)+, 1h and c, 4
*n=5, > 1 yr0032032−,5−,50c, 2−, 4
median age [range]:h and c, 3+, 1
3.6 yr [3-5.5]
PFIC2No.Portal TractSeptalCirrhosisFibrosisGiant HepatocytesNecrosisHCC DysplasiaLocalizationDuctular Reaction
*n=21, < 1 y49714125+, 21−, 15HCC, 1h, 1−, 12
median age [range]:+, 6 c, 11+, 9
3 mo [1-11] h and c, 9
*n=26, > 1 y0381501115−, 9−, 13HCC, 3h, 1−, 17
median age [range]:+, 17+, 13Dysplasia, 4c, 8+, 9
4.4 yr [1.1-13]h and c, 17

Normal-GGT PFIC Phenotype Without Identified Responsible Gene

In 10 patients with normal-GGT PFIC, belonging to nine families (PFIC? #1-9), ATP8B1 and ABCB11 sequencing did not show any deleterious abnormality (Supporting Table 1C). However, in nine of these patients heterozygous, homozygous, or compound heterozygous ATP8B1 and/or ABCB11 SNPs were identified (Supporting Tables 2, 3). The first symptoms and cholestasis appear within the first year of life (70% ≤3 months). PFIC? #9 patient developed LF at age 4 months and likely had PFIC2. Indeed, BSEP canalicular staining was negative (Supporting Fig. 1), sALT activity was 10× ULN, sAFP level 200× ULN, and liver histology at age 3 months showed numerous giant hepatocytes with severe lobular fibrosis. In other PFIC? children BESP canalicular staining performed in seven was normal and sALT and sAFP values available in three were ≤5× ULN and normal, respectively. Liver histology available in the first year of life showed mild fibrosis (n = 5) and a few giant hepatocytes in one patient (PFIC? #2). After 1 year, liver histology available in four patients showed cirrhosis. Among the eight patients who underwent BD or LT, one of them had exacerbation of diarrhea after successful BD (PFIC? #3). Biliary BA concentrations studied in PFIC? were either dramatically or slightly decreased (Table 6, PFIC? #2 [5.7 mmol/L]). Taken together, in PFIC? children signs at diagnosis or during disease course were suggestive of ATP8B1 or ABCB11 defects, but one or several features were not in concordance with the presumed defect, except in PFIC? #9.

Table 6. Outcome of External Biliary Diversion in 15 Children with Normal-GGT PFIC
 Age at Biliary DiversionBile Acids: InitialBile Acids: PostdiversionOutcome of Biliary Diversion (Duration of Success in Years)
Serum (μmol/L)Bile (mmol/L)Serum (μmol/L)Bile (mmol/L)
  • S, success; PF, partial failure; F, failure (see Materials and Methods); na, not available; PFIC?, PFIC of unknown cause.

  • *

    Patient harboring heterozygous BSEP mutation p.E297G.

  • Patient harboring homozygous BSEP mutation p.E297G.

PFIC1 no. 91.5 year251na265naF
PFIC2 no. 31.5 year2620.191850.41F
PFIC2 no. 94.5 year980.82.533.5S (13)
PFIC2 no. 10a5 year3440.104420.11F
PFIC2 no. 125 year1940.113320.18F
PFIC2 no. 1710.5 year2270.10470.9PF
PFIC2 no. 18*6 year4400.161250.5PF
PFIC2 no. 19*5 year3430.232330.44PF
PFIC2 no. 243 year3280.13258029F
PFIC2 no. 303.5 year5960.223900.70F
PFIC? no. 37.5 year3750.391.911S (10)
PFIC? no. 54.7 year262na420S (8)
PFIC? no. 63 year2804.14117S (2)
PFIC? no. 6   2011.2then PF
PFIC? no. 7a6 year256na1495PF
PFIC? no. 7b2 year1910.092770.50F

Response to UDCA Treatment and Outcome After BD or LT (Fig. 1; Table 3; Supporting Table 1)

PFIC1 Patients.

All 13 children received UDCA. A positive response to UDCA was observed in five patients. Among the eight other patients with intermediate or negative response to UDCA, BD was performed in one child but was unsuccessful (Table 6). All in all, six patients received a LT. In each case extrahepatic features were present after LT. In two patients a second LT was necessary between 16 and 22 years because of LF induced by massive steatosis. Unfortunately, diarrhea and steatosis recurred progressively. Four transplanted patients are alive and two died post-LT. No death related to liver disease occurred before LT.

Figure 1.

Outcome of 62 normal-GGT PFIC patients. Response to UDCA treatment and outcome after biliary diversion or liver transplantation in PFIC1 (A), PFIC2 (B), and PFIC? (C) patients. UDCA response: positive, normalization of serum liver tests; intermediate, partial improvement of serum liver tests; negative, no improvement of serum liver tests. PFIC?, PFIC of unknown cause; age: median [range] values.

PFIC2 Patients.

Two patients were lost during follow-up, one died of cerebral bleeding at age 4 month, and four other patients did not receive UDCA but underwent LT as first-line therapy. One of them died during LT. Thus, 32 children with PFIC2 received UDCA including the three patients initially presenting as BRIC2. Twelve patients had a positive response to UDCA, persisting in 10, but in two patients relapse under UDCA occurred at age 5 and 8.5 years, respectively. Among the eight patients presenting with a negative response, one died of LF at age 4 years. BD performed in nine patients has been successful in one for 13 years, but failed partially or totally in eight (Table 6). Among patients with unsuccessful UDCA treatment or BD, 11 were transplanted and four are waiting for a LT. No extrahepatic manifestation occurred in the 14 alive PFIC2 transplanted patients.

Normal-GGT PFIC Patients Without Identified Responsible Gene.

Among the 10 PFIC? children, two patients did not receive UDCA and underwent a LT. Eight children received UDCA. Two patients are alive with NLT under treatment and one had an initial positive response to UDCA but relapsed and underwent BD. One and four children had an intermediate and a negative response, respectively, and four of these patients underwent BD. All in all, five patients underwent BD: two had a successful outcome over an 8-10 year period, one a transient success for 2 years, one a partial failure, and one a failure (Table 6). In cases of unsuccessful UDCA or BD treatment, LT was performed in two patients and planned in one. None of the three alive transplanted patients exhibited extrahepatic signs post-LT. No death related to liver disease occurred before LT in PFIC? patients.

Genotype-Phenotype Correlation

Within each PFIC1 or PFIC2 phenotype we did not find an obvious genotype-phenotype correlation. Indeed, when we analyzed patients with biallelic mutations we were unable to correlate the types of mutations (missense versus truncating) with severity or presence of clinical or biochemical events. In addition, different phenotypes were observed between siblings sharing the same genotype (PFIC1 #4, PFIC2 #14). However, we found tendencies for a better response to UDCA or BD in a few children with missense mutations. Among the 15 children with PFIC1 or -2, alive with NLT with UDCA, 60% and 100% of PFIC1 and PFIC2, respectively, had at least one allele harboring a missense mutation. The only PFIC2 child with a total success of BD harbored homozygous mutation p.E297G and two PFIC2 children with a partial failure of BD were heterozygous for p.E297G mutation. In PFIC2, children with biallelic truncating mutations had a tendency to have lower biliary BA concentration (median: 0.02 mmol/L) and lower age at LT (median: 4.4 years) than those with one truncating and one missense mutation (0.13 mmol/L; 6.5 years), and than those with biallelic missense mutations (0.34 mmol/L; 7.2 years). Whatever the mutation type, BSEP canalicular staining was negative in all PFIC2 patients tested. Biallelic missense mutations were found in BRIC2 patients who progressively evolved toward PFIC2. Among heterozygous family members, only PFIC2 family members had liver symptoms: two developed cholestasis due to estroprogestative birth pill (PFIC2 #3,34), two intrahepatic cholestasis of pregnancy (PFIC2 #3,20), two chronic pruritus (PFIC2 #20,34), and one cholelithiasis (PFIC2 #34). One sister had BRIC2 with the same biallelic mutation as her PFIC2 brother (PFIC2 #34).


We studied a cohort of 62 patients with normal-GGT PFIC in whom genetic analysis of ATP8B1 and/or ABCB11 was performed. The results show that: (1) some clinical, biochemical, and histological features allow the differentiation of PFIC1 from PFIC2 at diagnosis and/or during the course of the diseases; (2) a small proportion of normal-GGT PFIC is likely not due to ATP8B1 or ABCB11 mutations. In addition, this study recapitulates the natural history of normal-GGT PFIC and outcome with therapeutic options. Although this study had a retrospective design, its strength is the fact that most patients were followed in a single center, with few missing data and variation in medical practices.

The diagnosis of PFIC1 and PFIC2 has been established on the basis of ATP8B1 and ABCB11 mutation identification, respectively. Mutations were considered as a disease-causing mutation on the basis of (1) previously reported mutation; (2) truncating mutation; (3) predicted in silico splicing effect; (4) and for PFIC2, association with low biliary BA concentration and/or negative BSEP canalicular immunostaining.2, 3, 5-12, 15, 18, 20, 23, 26 Five patients had ATP8B1 or ABCB11 genotypes fully composed of new missense mutations for which the disease-causing effects are likely, as expected from locations in protein and/or changes in amino acid class. In seven PFIC2 patients, only a single mutated allele was identified despite sequence analysis of coding exons and splice sites. Five of these patients also harbored the p.V444A variant, which may lower the BSEP expression level.11 Mutations located outside the sequenced regions, such as in regulatory domains, or microdeletions of one or more exons cannot be excluded.7 Such defects could explain the phenotype of the PFIC? patient who presented all features of PFIC2. In ≈15% of normal-GGT PFIC patients, ATP8B1 or ABCB11 SNPs were found but no obvious gene mutations were identified. Recently, some ABCB11 SNPs have been shown to possibly affect BSEP processing, function, or splicing.26 BSEP canalicular staining studied in six of our eight PFIC? patients harboring ABCB11 SNPs was positive, which is not in line with a defect of processing or splicing. We cannot exclude that mutations in genes other than ABCB11 or ATP8B1 are involved in PFIC? patients.1 In our study, phenotypic analysis did not allow us to identify features that could exclude PFIC1 or -2, and therefore that could justify prospectively suspending ATP8B1 or ABCB11 sequencing. We suggest that in normal-GGT PFIC, negative BSEP canalicular staining should lead to studying ABCB11 first, whereas normal BSEP canalicular staining would suggest studying ATP8B1 first.7, 9

The first symptoms did not differ substantially between PFIC1 and PFIC2. One PFIC1 child had diarrhea at presentation and two PFIC2 children had cerebral bleeding due to vitamin K deficiency. PFIC2 patients tended to develop cholestasis earlier than PFIC1 patients (within the first month of life in some cases). We found that initial serum levels of ALT and AFP, severe lobular lesions with numerous giant hepatocytes, early LF, cholelithiasis, HCC, very low biliary BA concentration, and negative BSEP canalicular staining suggest PFIC2, whereas an absence of these signs and/or presence of extrahepatic manifestations suggest PFIC1.1, 7-10, 13, 15-18 Clearly, PFIC2 has a severer course in infancy and carries the risk of early LF and HCC, both justifying close monitoring.7 This could be related to the major defect of biliary BA secretion observed in PFIC2 leading to hepatocyte BA overload and ongoing cell damage and regeneration.9, 15, 20, 26 The high frequency of cholelithiasis we showed in PFIC2 can also be related to the very low concentration of BA in bile promoting the formation of lithiasis. This study confirms that in PFIC1 the appearance or exacerbation of extrahepatic symptoms after LT may strongly hamper patient outcome and life quality, and that global outcome of PFIC2 after LT is good.8, 17, 27, 28 PFIC1 or PFIC2 phenotypes were not clearly correlated with mutation types. The difficulty to establish genotype-phenotype correlation may be partly due to the fact that some missense mutations (i.e., ABCB11) have been shown to possibly affect processing, function, or splicing.26 In PFIC2 this may explain why missense mutations are associated with negative BSEP canalicular staining and very low biliary BA concentration. Familial analysis showed that ABCB11 heterozygous status predisposes to develop cholestasis in certain circumstances, such as estroprogestative treatment or pregnancy.11, 12

In terms of treatment strategy, our policy has been to use UDCA as a first-line therapy as soon as possible and to discuss BD or LT depending on the outcome with UDCA.1, 19, 28 The data presented here show that about one-third of PFIC1, PFIC2, or PFIC? patients may benefit from UDCA therapy and that some of them reach adolescence with UDCA (PFIC1 #4a; PFIC2 #11,27,34,36; PFIC? #8). This response rate is higher than those reported in other series and could be related to the use of higher doses of UDCA, calculated according to body surface.14, 19, 28 Among the three patients in whom BD was successful with a normal and sustained normalization of BA concentration in bile and serum, one (PFIC2 #9) had severe lobular and portal fibrosis and the two others with PFIC? had mild lobular and portal fibrosis at time of BD. This lower BD response rate than in other reports might be due to a more advanced stage of liver fibrosis before BD was performed, because an absence of cirrhosis was previously thought to be predictive of success, and/or to how BD success or failure was defined.14, 22, 28-30 For example, pruritus relief was considered a BD success in a recent series and likely equals BD partial failure in our series.30ABCB11 genotype could also have a predictive value because the PFIC2 patient with BD success had a homozygous p.E297G mutation and two other PFIC2 patients with BD partial failure harbored on one allele the p.E297G mutation.31 This mutant is known to be associated with impaired membrane trafficking but to retain some transport activity when correctly targeted.32 The mechanism involved in the recovery of normal BA secretion after BD in patients harboring this mutant protein is unknown but could be related to canalicular targeting of the mutated BSEP. In this series, the combination of UDCA, BD, and LT allowed 87% of normal-GGT PFIC patients to be alive at a median age of 10.5 years. Around 50% of them did not require LT so far. Prospectively, it seems reasonable to propose that in case of refractory pruritus to medical treatment, and in the absence of LF, BD is performed before LT, especially in PFIC1 patients in whom LT has a relatively high morbidity/mortality.8, 17, 19, 22, 27-30

We evaluated clinical, biochemical, and histological features at presentation and during evolution in a large cohort of children with genetically characterized normal-GGT PFIC. This original study highlights the fact that PFIC1 differs from PFIC2 and recapitulates their outcomes. Only PFIC1 patients present obvious signs of extrahepatic disease that do not resolve after LT. PFIC2 patients present with evidence of greater hepatocellular injury and very severely impaired biliary BA secretion. These findings may be helpful to facilitate orientation of a specific immunohistochemical and molecular diagnosis in a child presenting with normal-GGT PFIC and suggest directions for prospective therapeutic approaches in selected patients.


We thank Dr. M. Hadchouel (INSERM and Hôpital Bicêtre, Paris, France), Dr. D. Habes, Dr. B Hermeziu, Dr. J. Cohen, and Dr. D. Debray (Hôpital Bicêtre, Paris, France), for their help in the follow-up of patients. We thank Drs. S. Fournier-Favre (Montpellier), E. Sokal (Bruxelles), J.P. Tholon (Reims), A. Jobert (Nantes), A. Lachaux, C. Rivet, and M.P. Cordier (Lyon), B. Roquelaure and S. Sigaudy (Marseille), A. Dabadie and L. Bridoux-Henno (Rennes), S. Danner, D. Astruc, Y. Alembik (Strasbourg), S. Thirion (Mulhouse), B. Descos (Nice), C. Dubuisson, B. Dubern (Paris), B Retali (Pontoise), B. Guillois (Caen), G. Maggiore (Pisa) for referring patients and/or biological samples to Bicêtre Hospital and/or for contribution of patient data. We also thank families who participated to the study.