Autoimmune Hepatitis/Sclerosing Cholangitis Overlap Syndrome in Childhood: A 16-Year Prospective Study

Transcription

1 Autoimmune Hepatitis/Sclerosing Cholangitis Overlap Syndrome in Childhood: A 16-Year Prospective Study GERMANA V. GREGORIO, 1 BERNARD PORTMANN, 2 JOHN KARANI, 3 PHIL HARRISON, 2 PETER T. DONALDSON, 2 DIEGO VERGANI, 4 AND GIORGINA MIELI-VERGANI 1 Abbreviations: AIH, autoimmune hepatitis; IgG, immunoglobulin G; ANA, anti-nuclear antibody; SMA, anti-smooth muscle antibody; ASC, autoimmune sclerosing cholangitis; LKM1, anti-liver kidney microsomal type 1 antibody; PSC, primary sclerosing cholangitis; ERCP, endoscopic retrograde cholangiopancreatography; ANCA, antineutrophil cytoplasmic antibody; UDCA, ursodeoxycholic acid; AST, aspartate transaminase; AP, alkaline phosphatase; GGT, -glutamyl transpeptidase; OR, odds ratio; INR, international normalized prothrombin ratio. From the Department of 1 Child Health, 2 Institute of Liver Studies, and 3 Department of Radiology, King s College Hospital, London, England; and 4 Institute of Hepatology, University College London, London, England. Received September 6, 2000; accepted December 4, Dr. Donaldson s current address is: Centre for Liver Research, School of Clinical Medical Sciences, University of Newcastle, Newcastle-upon-Tyne, England. The data in this article were presented in part in J Hepatol 2000;32(Suppl 2):7-8. Supported by the Children s Liver Disease Foundation, Birmingham, UK (to G.M.-V.). Address reprint requests to: Professor Giorgina Mieli-Vergani, Department of Child Health, King s College Hospital, Denmark Hill, London SE5 9RS, England. giorgina.vergani@kcl.ac.uk; fax: Copyright 2001 by the American Association for the Study of Liver Diseases /01/ $35.00/0 doi: /jhep To investigate whether sclerosing cholangitis with an autoimmune serology characteristic of autoimmune hepatitis (AIH) and AIH are distinct entities, we studied 55 consecutive children with clinical and/or biochemical evidence of liver disease and circulating antinuclear (ANA), anti-smooth muscle (SMA), and/or liver-kidney-microsomal type 1 (LKM1) autoantibodies. They underwent liver biopsy, direct cholangiography, sigmoidoscopy, and rectal biopsy at presentation. Twenty-eight were diagnosed as AIH in the absence and 27 autoimmune sclerosing cholangitis (ASC) in the presence of radiological features of cholangiopathy. Twenty-six ASC and 20 AIH had ANA and/or SMA; 1 ASC and 8 AIH LKM1 autoantibody. Similarities between the 2 conditions included most clinical and biochemical parameters and a lower frequency of HLA DR4. Inflammatory bowel disease and histological biliary changes were more common in ASC; coagulopathy, hypoalbuminemia, lymphocytic periportal hepatitis, and HLA DR3 were more common in AIH. Histological biliary changes were observed in 65% of ASC and 31% of AIH patients. Eighty-nine percent responded to immunosuppression. Follow-up liver biopsies from 17 ASC and 18 AIH patients had similarly reduced inflammatory activity and no progression to cirrhosis. Sixteen follow-up cholangiograms from AIH patients and 9 from ASC patients were unchanged, while 8 ASC patients showed a progressive cholangiopathy. One child with AIH and ulcerative colitis developed sclerosing cholangitis 8 years after presentation. At 2 to 16 years (median, 7 years) from presentation, all patients are alive, including 4 ASC patients who underwent liver transplantation. In conclusion, ASC and AIH are similarly prevalent in childhood; cholangiography is often needed to distinguish between these 2 entities, which are likely to lie within the same disease process. (HEPATOLOGY 2001;33: ) Sclerosing cholangitis is a progressive hepatobiliary disease of unknown etiology characterized by chronic inflammation and fibrosis of the intra- and/or extrahepatic bile ducts, with obliteration of peripheral ducts and cholangiographic evidence of stricturing and dilatation of all or parts of the biliary tree. 1 An overlapping syndrome between autoimmune hepatitis (AIH) and sclerosing cholangitis has been anecdotally reported on several occasions both in children 2-4 and adults. 5-8 A retrospective study 4 from our Unit on 13 children with sclerosing cholangitis showed that in 40% of the cases, the clinical, biochemical, immunological, and histological presentation was indistinguishable from that of AIH. In both conditions, there was an increase in the levels of immunoglobulin G (IgG) and presence of circulating non organ-specific nuclear (ANA) and/or smooth muscle (SMA) autoantibodies. Histological abnormalities common to both diseases were inflammatory portal tract infiltrate and interface hepatitis. Most of the reported overlapping cases 4-8 were originally diagnosed as AIH, the diagnosis of sclerosing cholangitis having been made years later after demonstration of characteristic bile duct changes on cholangiography, which was performed because of the development of biliary features on follow-up liver biopsy. This has been interpreted as evolution from AIH to sclerosing cholangitis, although cholangiographic studies excluding sclerosing cholangitis at presentation were not available. To investigate whether autoantibody-positive sclerosing cholangitis ( autoimmune sclerosing cholangitis [ASC]) and autoimmune hepatitis in childhood are distinct entities at onset and at follow-up, we have conducted a prospective study from 1984 to During this time, 55 consecutive children presenting with clinical and/or biochemical evidence of liver disease and a serological profile compatible with AIH underwent liver biopsy, cholangiography, sigmoidoscopy, and rectal biopsy at the earliest possible time after presentation. Follow-up liver biopsies and cholangiographic studies were performed in most patients to monitor disease progression over a median follow-up period of 7 years. The clinical course and outcome of these patients are described. 544 PATIENTS AND METHODS A total of 76 children referred to our Unit between January 1984 and January 1997 had clinical and/or biochemical evidence of liver disease associated with circulating autoantibodies. All known causes of chronic liver diseases in childhood were investigated as previously

2 HEPATOLOGY Vol. 33, No. 3, 2001 GREGORIO ET AL. 545 described. 9 Because the diagnosis of sclerosing cholangitis for the purpose of this study was based on the demonstration of specific cholangiographic changes, 1 21 patients, too ill to undergo cholangiography, were excluded. An additional case was excluded because of inadequate cholangiographic visualization of the intrahepatic bile ducts to third-order level. Cholangiographic studies were satisfactory in 55 patients. In 54, a sigmoidoscopy and rectal biopsy were performed to assess the presence of an associated distal colitis. Informed consent for all procedures was obtained from patients and/or their parents. One patient refused sigmoidoscopy. The 55 patients included in the study were followed up through December 31, During the period of time the 55 patients were recruited (January 1984 January 1997), 22 children with sclerosing cholangitis resulting from other causes were diagnosed: 6 had primary immunodeficiency, 2 had Langerhans cell histiocytosis, and 5 had neonatal sclerosing cholangitis, while 9, with no identifiable etiology and negative for ANA, SMA, and liver kidney microsomal type 1 (LKM1) autoantibodies, were classified as primary sclerosing cholangitis (PSC). The clinical notes of these 9 children were reviewed retrospectively. Cholangiography. Sclerosing cholangitis was defined as the presence of irregularity, stricturing, focal dilatation, or parallelism within the intra- and/or the extrahepatic bile ducts as demonstrated at direct cholangiography. 10 Endoscopic retrograde cholangiopancreatography (ERCP) was performed in 51 patients, percutaneous transhepatic cholangiography in 3, and operative cholangiography in 1. Forty-eight cholangiograms were available for retrospective review by 2 independent reviewers (J.K. and P.H.), who were unaware of the clinical and histological findings, and were scored as previously described. 10 When there was disagreement between the 2 observers on the presence of cholangiopathy, a diagnosis of sclerosing cholangitis was excluded. Histology. Liver specimens for baseline histological examination were obtained by percutaneous needle biopsy in 53 patients, by transjugular route in 1, and during surgery in 1. All follow-up biopsies were obtained percutaneously. A single pathologist (B.P.) who was blinded to the patients clinical details and chronological order of the biopsies reviewed the histology retrospectively under code. The severity of the necroinflammatory process was semiquantitatively graded on a scale of 0 to 4 (0 absent; 1 minimal; 2 mild; 3 moderate; 4 severe), and separate scores were obtained for each of the following 11 : portal tract inflammation, lobular/nodular activity, and interface (piecemeal) necrosis, with the latter defined as biliary and/or lymphocytic type. 12 A histological inflammatory activity index was derived from the sum of the 3 separate scores for the different components of the necroinflammatory lesions (maximal score 12). The degree of fibrosis was recorded as follows: 0, no fibrosis; 1, enlarged fibrotic portal tracts; 2, periportal fibrosis (radiating/nonbridging septa); 3, septal bridging fibrosis; and 4, definite cirrhosis. The presence or absence of acute and/or chronic cholangitis, multinucleated giant cells, lymphocytic aggregates, and copperbinding protein was also recorded. Autoantibodies. Tissue autoantibodies were tested at a starting dilution of 1/10 in phosphate-buffered saline by indirect immunofluorescence on rat liver, kidney, and stomach. 13 All positive sera were titered to extinction. Antineutrophil cytoplasmic antibodies (ANCA) and antibodies to cytochrome P4502D6 14 were determined by indirect immunofluorescence on ethanol-fixed and formalin-fixed human neutrophils and radioligand immunoassay, 14 respectively, and the results were reported as positive or negative. HLA Typing. HLA typing was performed in 45 of the 50 white patients and in 1 mixed-race child (white/west Indian) and compared with 134 white control subjects. HLA A and B were typed by microcytotoxicity 15 using sera provided by the United Kingdom Transplant Service Special Authority. Antigen assignments were made for broad antigens only. 16 HLA genotyping for 24 alleles of DR and 15 alleles of DQB was performed as previously described. 16 Treatment. Immunosuppressive treatment was given according to the protocol used in our Unit for autoimmune hepatitis. 9 After 1992, following reports of a beneficial effect of ursodeoxycholic acid (UDCA) in adult PSC and of its possible immunomodulatory properties, we have added this drug to the treatment of most patients diagnosed as ASC. Statistical Analysis. All data are expressed as median and range. Comparison between categorical values was performed using 2 or Fisher s exact tests as appropriate. The results of the HLA data are given as odds ratios. 20 The Bonferroni s correction for multiple testing was not applied, because only known class II associations are presented. Survival analyses were performed by the Kaplan-Meier product-limit method. The starting time for the analyses was the date at presentation to our Unit. Patients who underwent transplantation were censored on the date of transplantation. P.05 was considered significant. RESULTS Diagnosis. All 55 patients were autoantibody-positive (see Laboratory Investigations) and underwent cholangiography from 3 days to 12 months (median, 1.1 month) after presentation to our Unit. Forty-eight cholangiographic films were available for retrospective independent review. There was agreement between the 2 observers in 45 cases, changes consistent with sclerosing cholangitis being present in 23. Three cases for whom there was no agreement were considered as not fulfilling the radiological diagnosis of sclerosing cholangitis. In 7 patients for whom the films were not available for review, the original radiological diagnosis (J.K.) was adopted, sclerosing cholangitis being diagnosed in 2 patients. A cholangiographic diagnosis of sclerosing cholangitis was made in 27 (49%) patients, who henceforth are referred to as ASC. Eighteen (67%) of them had intra- and extrahepatic bile duct abnormalities, and 9 (33%) had abnormalities confined to the intrahepatic bile ducts. The radiological score of 23 baseline cholangiographic studies from the ASC patients available for review is detailed in Table 1. The 28 (51%) patients with no radiological signs of cholangiopathy were diagnosed as AIH, and all fulfilled internationally agreed diagnostic criteria for this condition. 21 Clinical Features. Clinical features at presentation are summarized in Table 2. Nine of the 10 ASC patients with no history of jaundice were referred for investigation following the discovery of hepatomegaly and/or abnormal liver function tests during consultation for inflammatory bowel disease (4), attention deficit disorder (1), abdominal pain (2), fatigue, anorexia, and weight loss (1), and urticarial rash (1); 1 with TABLE 1. Cholangiographic Findings at Diagnosis of 23 Patients With ASC Common bile duct Maximal caliber (mm), median (range) 4.4 (2-9) Presence of stricture 2 (9%) Mucosal irregularity 15 (65%) Extrahepatic bile duct score 0, absent 6 (26%) 1, mild 10 (44%) 2, moderate 6 (26%) 3 and 4, severe 1 (4%) Intrahepatic bile duct score 0, absent 0 1, mild 9 (39%) 2, moderate 9 (39%) 3 and 4, severe 5 (22%) Presence of calculi in common bile duct, common hepatic duct, cystic duct and gallbladder 2 (9%) Presence of abnormal pancreatic duct 3 (13%)

3 546 GREGORIO ET AL. HEPATOLOGY March 2001 TABLE 2. Baseline Demographic and Clinical Features in 27 Patients With ASC and 28 With AIH ASC (n 27) AIH (n 28) P Age at diagnosis (yr), median (range) 11.8 (2.3-16) 10.5 (2.2-14).16 Female 15 (55%) 22 (79%).12 Mode of presentation, n (%) Prolonged acute hepatitis 10 (37%) 14 (50%).6 Chronic hepatitis 7 (26%) 5 (18%) No history of jaundice 10 (37%) 9 (32%) Signs at diagnosis Jaundice 15 (56%) 19 (68%) Hepatomegaly 15 (56%) 19 (68%) Splenomegaly 14 (52%) 17 (61%) Hepatosplenomegaly 9 (33%) 12 (43%) Spider nevi 5 (18%) 7 (25%) Variceal bleeding 0 (0%) 2 (7%) Ascites 2 (7%) 4 (25%) Symptoms at diagnosis Diarrhea 10 (37%) 8 (29%) Abdominal pain 8 (30%) 8 (29%) Anorexia 5 (18%) 8 (29%) Fatigue 6 (22%) 8 (29%) Vomiting 5 (18%) 9 (32%) Pruritus 2 (7%) 7 (25%).08 Weight loss 3 (11%) 2 (7%) Associated autoimmune disorders* 13 (48%) 11 (39%) Inflammatory bowel disease 12 (44%) 5 (18%).03 Ulcerative colitis 5 2 Indeterminate colitis 4 1 Crohn s disease 3 1 Acute proctitis 0 1 Autoimmune thyroiditis 0 4 (14%) Insulin-dependent diabetes mellitus 0 2 (7%) Urticaria pigmentosa 0 1 (4%) Hypoparathyroidism and Addison s disease 0 1 (4%) Thrombocytopaenia 1 0 Autoimmune disorders in first degree relatives* 10 (37%) 20 (71%).02 * Some subjects have more than one disorder. splenomegaly but no hepatomegaly and normal liver function tests was investigated for recurrent pain over the right iliac fossa. Among the 9 AIH patients with no history of jaundice, 2 were brought to medical attention because of bleeding from esophageal varices, and 7 were found to have hepatomegaly and/or abnormal liver function tests during consultation for inflammatory bowel disease (2), insulin-dependent diabetes mellitus (1), recurrent bruising (1), and symptoms of fatigue, anorexia, and weight loss (3). Laboratory Investigations. Laboratory investigations at presentation are summarized in Table 3. Among the 27 patients with ASC, 6 were positive for ANA alone, 6 for SMA alone, 14 for both SMA and ANA, and 1 for LKM1. Among the 28 patients with AIH, 6 were positive for ANA alone; 2 for SMA alone; 10 for ANA and SMA; 6 for LKM1; 1 for LKM1 and ANA; and 1 for LKM1, ANA, and SMA. Two AIH patients, 1 with cirrhosis on initial biopsy and 1 with associated juvenile rheumatoid arthritis, were autoantibody-negative at presentation, but became positive during treatment (ANA/SMA after 3 months and SMA after 1 month, respectively). Overall, ANA and/or SMA positivity was observed in the same proportion of patients with ASC (26 of 27) or AIH (20 of 28), while LKM1 antibody was significantly more common in patients with AIH (8 of 28 vs. 1 of 27; P.014). ANCA, positive in 30 patients (20 ASC), had a perinuclear pattern in 29 children and a cytoplasmic pattern in 1 child with ASC. Only 1 of the 9 LKM1-positive patients was positive for ANCA, and this was a child with AIH also positive for ANA. There was no association between the presence of ANCA and inflammatory bowel disease, 22 ANCA being positive in 13 of 17 patients with, but also in 16 of 29 without, inflammatory bowel disease. Six (22%) patients with ASC and 2 (7%) with AIH had normal aspartate transaminase (AST) ( 50 IU/L), but even excluding these patients, the median levels of AST at presentation was higher in AIH (ASC: 106 IU/L [51-1,215] vs. AIH: 358 IU/L [83-4,830]; P.007). Sixteen (59%) patients with ASC and 13 (46%) with AIH had normal alkaline phosphatase (AP) ( 350 IU/L) and 8 (30%) with ASC and 7 (25%) with AIH had normal -glutamyl transpeptidase (GGT) ( 50 IU/L) activity. Combined normal baseline AP and GGT levels were similarly observed in both groups (ASC: 7 [26%] vs. AIH: 4 [14%]). A ratio of more than 3 between the absolute values of AP and AST was significantly more common in children with ASC. HLA Type. Forty-six (22 ASC) patients were typed for HLA DR. When compared with controls, AIH patients had a significantly higher frequency of DR3 (15/24 vs. 35/134; P.0004; odds ratio [OR] 4.71) and a significantly lower frequency of DR4 (4/24 vs. 54/134; P.027; OR 0.3). DR15 was less frequent in AIH patients than in controls, but not significantly (1/24 vs. 27/134; not significant; OR 0.17), and DR13 was similarly frequent in patients and controls (6/24 vs. 26/134; TABLE 3. Baseline Laboratory Features in 27 Patients With ASC and 28 With AIH ASC (n 27) AIH (n 28) P Total bilirubin, mol/l (nv: 20) 20 (4-178) 35 (4-306).040 AST, IU/L (nv: 50) 102 (18-1,215) 333 (24-4,830).002 GGT, IU/L (nv: 50) 129 (13-948) 76 (29-383) NS n 26 AP, IU/L (nv: 350) 303 (104-1,710) 356 ( ) NS AP/AST ratio 3.96 ( ) 1.14 ( ).0013 Total protein, g/l (nv: 60-80) 77 (56-104) 76 (49-117) NS n 26 Albumin, g/l (nv: 30-50) 39 (27-54) 35 (25-47).051 INR (nv: ) 1.1 ( ) 1.2 ( ).009 IgG, g/l (nv: 5-18) 19.4 ( ) 24 ( ) NS IgA, g/l (nv: ) 1.5 ( ) 1.5 ( ) NS IgM, g/l (nv: 0.5-2) 1.6 ( ) 1.7 ( ) NS C3, g/l (nv: ) 1.1 ( ) 0.86 ( ) NS C4, g/l (nv: ) 0.18 ( ) 0.14 ( ) NS Reciprocal of ANA titer 160 (20-20,480) 160 (10-2,560) NS n 20* n 18* Reciprocal of SMA titer 160 (10-1,280) 160 (10-1,280) NS n 20* n 13* Reciprocal of LKM1 titer (40-10,240) NA n 1* n 8* Percentage ANCA-positive 74% 36%.009 NOTE. Values expressed as median (range). Abbreviations: nv, normal value; NA, not applicable; NS, not significant. * Number of patients positive at presentation. All LKM1-positive patients were reactive to eukaryotically expressed CYP2D6. Number of patients tested.

4 HEPATOLOGY Vol. 33, No. 3, 2001 GREGORIO ET AL. 547 not significant). Of the 15 AIH patients who were DR3-positive, 13 were B8-positive and 11 were A1-positive. All the 13 DR3-positive patients typed for DQ were DQB1*0201. Of the 4 DR4-positive patients, 2 were DQB1*0302. When compared with controls, ASC patients had a significantly lower frequency of DR4 (4/22 vs. 54/134; P.047; OR 0.33), but a similar frequency of DR3 (8/22 vs. 35/134; NS; OR 1.62), DR15 (1/22 vs. 27/134; not significant; OR 0.19), and DR13 (7/22 vs. 26/134; not significant; OR 1.94). Of the 8 ASC patients positive for DR3, 7 were B8-positive, 6 were A1-positive, and 7 were DQB1*0201-positive. Of the 4 positive for DR4, 2 were DQB1*0302. Only 1 (5%) ASC patient was positive for DRB1*1501 compared with 27 (20%) controls, but this difference was not significant, despite an OR 0.19, because of the small patient sample. DR3 was more frequent in AIH than in ASC, though the difference did not reach a conventional level of statistical significance (15/24 vs. 8/22; P.07). Histological Findings. Histological findings are summarized in Table 4. Liver biopsy was performed before treatment in 39 patients and after commencement of immunosuppressive therapy in 16 patients (7 ASC). In 12 (4 ASC) of these 16 patients, percutaneous needle biopsy was possible from 2 weeks to 18 months (median, 4 months) after starting treatment when the international normalized prothrombin ratio (INR) normalized, while 4 patients (3 ASC) were receiving prednisolone for associated inflammatory bowel disease at referral and had been treated for 5 to 9 months before the TABLE 4. Histological Features on Initial Biopsy in 26 Patients With ASC and 26 With AIH ASC (n 26) AIH (n 26) P Moderate/severe: Portal tract inflammation 15 (58%) 24 (92%).004 Periportal hepatitis 9 (35%) 15 (58%).082 Lobular activity 8 (31%) 16 (61%).025 Histological inflammatory activity index, median (range) 5 (1-11) 8 (3-12).004 Periportal hepatitis (any degree) n 19 n 26 Biliary 8 (31%) 4 (15%).049 Lymphocytic 8 (31%) 17 (65%) NS Mixed 3 (16%) 5 (19%) NS Histological stage 1 enlarged fibrotic portal tract 4 (15%) 3 (12%) NS 2 periportal fibrosis 5 (19%) 5 (19%) NS 3 septal and/or bridging fibrosis 13 (50%) 12 (46%) NS 4 definite cirrhosis 4 (15%) 6 (23%) NS Acute and/or chronic cholangitis 9 (35%) 3 (12%).049 Multinucleated giant cells 3 (12%) 3 (12%) NS Lymphocytic aggregates 7 (27%) 7 (27%) NS Copper-binding protein 15 (58%) 13 (65%) NS n 22 Histological diagnosis Sclerosing cholangitis 11 (42%) 1 (4%) Chronic hepatitis 6 (23%) 17 (65%) Chronic hepatitis with biliary features* 5 (19%) 7 (27%) Minimal portal inflammation and fibrosis 3 (12%) 1 (4%) Biliary cirrhosis 1 (4%) 0 Abbreviation: NS, not significant. * Bile duct damage, acute and/or chronic cholangitis, biliary periportal hepatitis. biopsy was performed. Three patients, 2 (1 with ASC) who underwent liver biopsy more than 1 year after starting treatment and 1 (AIH) in whom the specimen showed cirrhosis but was too small for further evaluation, were excluded from the analysis of histological data. Only 2 patients, both with ASC, had the typical onion-skin appearance of periductular fibrosis. Four patients (3 ASC) had only minimal inflammatory changes. None of them had been previously treated with prednisolone, but 1 was on salazopyrine for inflammatory bowel disease. Treatment. A total of 51 children (23 ASC) were treated with immunosuppression, 49 from onset and 2 during followup. Of the 28 children with AIH, 26 were treated with prednisolone at a starting dose of 2 mg/kg/d (maximum of 60 mg/d), rapidly decreased to the minimal dose able to maintain normal transaminases (2.5-5 mg/d). 9 Twenty-four of these children required addition of azathioprine (1 to 2 mg/kg/d) because of increasing AST levels on tapered dose of steroids (15) or because of steroid side effects (9). One child, already on steroids at referral, had normal liver function tests and was kept on prednisolone only, while the remaining child, who at diagnosis had normal AST (36 IU/L) but IgG of 25.9 g/l, LKM1 1/1,280, and moderate inflammatory activity on liver biopsy, was given azathioprine (1 mg/kg/d) alone, because she also had insulin-dependent diabetes mellitus. This child, however, required addition of prednisolone 4 years later because of abnormal AST. Of the 27 children with ASC, 23 were treated with immunosuppression, 21 from diagnosis and 2 during follow-up. These 2 children had been treated with UDCA alone at presentation because of mild histological changes, but required immunosuppression later, 1 because of symptomatic inflammatory bowel disease, and the other because of increased AST. Of 18 children with abnormal AST treated with steroids, 16 required addition of azathioprine because of increasing AST levels on tapered dose of steroids (11) or because of steroid side effects (5). In 6 patients in whom there was no normalization of AST levels 3 months to 4 years (median, 3 years) from presentation, penicillamine (3 patients, 20 mg/kg/d), cyclosporin A (2 patients, 1-2 mg/kg/d), and colchicine (1 patient, 25 g/kg/d) were added to the treatment. In 10 children, UDCA was added from 3 months to 12 years (median, 7 months) after starting immunosuppressive treatment. Four children did not receive immunosuppression. Of these, 3 were treated with UDCA, 2 having normal AST and mild inflammatory changes on liver biopsy, and 1 because of mucus glandular hyperplasia and dysplasia of the bile ducts, suggesting premalignant changes. The remaining child, who had normal AST, IgG 13.6 g/l, SMA 1/160, mild inflammatory changes on liver biopsy, and a cholangiopathy involving the extrahepatic bile duct with some irregularity of the first order, but normal second- and third-order intrahepatic ducts, underwent cholecystectomy for chronic hypertrophic cholecystitis and did not receive medical treatment. Seventeen (12 ASC) patients were treated with salazopyrine (250 to 750 mg/d) for an associated inflammatory bowel disease, including 1 patient with ASC who was clinically symptomatic but refused a rectal biopsy. Biochemical and Clinical Follow-up. Median follow-up time was 7 (2-16) years (ASC: 6 [2-16] vs. AIH: 8 [3-15]; P.17). Biochemical and clinical response to immunosuppressive treatment was satisfactory in both groups. All patients with

5 548 GREGORIO ET AL. HEPATOLOGY March 2001 AIH who had abnormal AST, bilirubin, GGT, AP, or albumin at presentation achieved normal values at follow-up. Only 1 child with AIH continued to have mildly increased INR. Of those patients with ASC who had abnormal values at presentation, during follow-up, 83% normalized AST, 73% bilirubin, 89% GGT, 92% AP, 89% albumin, and 100% INR. Time to normalization of AP, albumin, and INR was significantly shorter in ASC than in AIH (Table 5). Three patients with ASC continued to have abnormal liver function tests on immunosuppressive treatment. During the period of observation, 9 (33%) patients with ASC and 10 (36%) with AIH had 1 or more episodes of biochemical relapse (defined as at least 2-fold increase in AST) and responded to a temporary increase in the prednisolone dose. During relapse, 60% of the AIH patients and 75% of the ASC patients had increased GGT, and 30% of the AIH patients and 25% of the ASC patients had increased AP. The patient not treated with immunosuppression because of the presence of biliary epithelial dysplasia on histology showed AST normalization and a 68% reduction in GGT levels (645 to 208 IU/L) after 1 year of UDCA treatment. Discontinuation of treatment was attempted in 9 children (2 ASC) who had normal liver function tests for at least 1 year and no necroinflammatory activity in a follow-up liver biopsy. Four (3 AIH) were able to discontinue therapy successfully (2, 3, 3, and 5 years after starting immunosuppression); 3 (all ANA/SMA AIH) have stopped prednisolone and are in the process of decreasing azathioprine; and 2 patients (1 ASC and 1 LKM1 AIH) relapsed 6 and 8 months after stopping treatment but went in remission when prednisolone was restarted. A further 2 patients with ASC discontinued their treatment without medical advice 2 and 4 years after starting immunosuppression, and have remained in remission 4 and 7 years later. No follow-up liver biopsies or ERCPs were performed in these patients after discontinuation of treatment. Follow-up Liver Biopsies. Sixty-eight follow-up liver biopsies were obtained in 35 (17 ASC) patients (median of 2 [1-6] biopsies per patient) from 9 to 156 (median, 48) months after the initial biopsy. A significant decrease in the histological inflammatory activity index score was similarly observed in the patients with ASC (initial: 7 [2-11] vs. final: 2 [1-6]; P.003) and AIH (initial: 10 [3-12] vs. final: 2 [0-6]; P.001). A decrease in inflammatory activity was also observed in the follow-up liver biopsies taken 1, 2, and 4 years from presentation in the 3 children treated with UDCA alone. Progression to cirrhosis was not observed in any patient, but 1 with ASC showed disappearance of bile ducts. A patient with ASC who had no biliary features on initial biopsy showed pericholangitis with bile duct changes in a liver biopsy performed 2 years later. Four patients with ASC who had cholangitic changes on TABLE 5. Time to Normalization of Biochemical Parameters in Patients With Baseline Abnormal Values Treated With Immunosuppression ASC (months) AIH (months) P AST (15 ASC, 26 AIH) 2.1 ( ) 6 (0.8-25).118 GGT (17 ASC, 20 AIH) 7 (1-96) 12 (1-40).921 AP (11 ASC, 17 AIH) 2 (1-40) 5 (1-48).015 Bilirubin (8 ASC, 20 AIH) 1.5 (0.5-12) 1.25 ( ).731 Albumin (6 ASC, 12 AIH) 0.5 (0.5-1) 2 (0.25-8).016 INR (8 ASC, 12 AIH) 1 (1-6) 6 (0.5-15).021 NOTE. Results are expressed as median (range). initial biopsy showed no biliary features on repeat biopsies taken 2, 4, 7, and 9 years later. Four of 7 patients with AIH who had cholangitic changes at presentation had no biliary features 3, 4, 5, and 7 years later, while 3 had persistent biliary changes 3, 4, and 7 years after the initial biopsy, including 1 LKM1-positive girl whose baseline liver biopsy showed changes characteristic of sclerosing cholangitis, but whose ERCP did not show cholangiopathy. Follow-up Cholangiographies. A total of 36 follow-up ERCPs were performed in 34 patients (17 ASC) from 1 to 9 years (median, 5 years) after the initial cholangiography. Two children had 2 follow-up ERCPs each, 4 and 6, and 5 and 6 years after the initial investigation. Of the 17 patients with ASC, 9, including 2 treated with UDCA only, had static disease, while 8 showed progression of intrahepatic and extrahepatic bile duct abnormalities, including 2 patients who had only intrahepatic bile duct changes initially. The median intra-/extrahepatic cumulative cholangiographic score for these 8 patients at presentation was 4 (1-5), and at follow-up was 5.5 (2-6). Follow-up ERCPs were performed at a median of 5 years (2.5-9 years) after baseline cholangiography in the patients with progressive disease, and at a median of 4 years (1-5 years) in those with no progression. In 7 of the 8 patients with worsening bile duct disease on ERCP, liver biopsies showed progression from periportal to bridging fibrosis in 2, same stage of fibrosis in 4, and ductopenia in 1. Liver histology in 6 patients whose follow-up ERCP showed no progression was similar to baseline in 5 and showed improved fibrosis in 1. Of the 17 patients with AIH who had follow-up ERCPs, 16 had no evidence of cholangiopathy after a median of 6 (1-9) years from the initial imaging, while 1 showed progression from a normal ERCP (Fig. 1A) to advanced intra- and extrahepatic bile duct disease 8 years after presentation (Fig. 1B). This patient is an ANA/SMA, HLA DR3 positive girl with ulcerative colitis and urticaria pigmentosa who normalized her AST 1 month after starting steroids and maintained normal liver function tests during the 8-year follow-up on prednisolone (2.5 mg alternate day) treatment. Her initial liver biopsy was reported as having features of chronic hepatitis with mild cholangitis, though subsequent biopsies performed after 1, 3, and 4 years during treatment were not suggestive of sclerosing cholangitis. A biopsy performed after the radiological detection of cholangiopathy was also not diagnostic of sclerosing cholangitis. Of a total of the 91 cholangiograhic studies performed, complications were observed in 7: 6 patients developed abdominal pain and increased amylase levels that resolved with conservative treatment within 24 hours, while 1 asymptomatic patient had evidence of mild periductular bleeding in the liver biopsy performed immediately after the ERCP. 23 Outcome. At the end of December 1999, all 55 patients were alive, including 4 after liver transplantation (all with ASC: 3, 1 with ductopenia, who never had normal AST, and 1 who normalized the AST but had progression of bile duct damage on follow-up ERCP). Transplant-free survival was longer in AIH than in ASC, the estimated survival rate without transplantation at 10 years being 65% for patients with ASC and 100% for patients with AIH (P.053). Forty-two (17 ASC) patients were taking immunosuppressive drugs, 6 (3 ASC) were on no treatment, and 3 (all ASC) were taking UDCA only. Fifteen (56%) patients with ASC and 24 (86%) with AIH had normal AST, GGT, AP, albumin, bilirubin, and INR.

6 HEPATOLOGY Vol. 33, No. 3, 2001 GREGORIO ET AL. 549 FIG. 1. (A) Endoscopic cholangiography performed at presentation in a patient with AIH, demonstrating normal intra- and extrahepatic bile ducts. (B) Endoscopic cholangiography 8 years later demonstrating extensive cholangiopathy affecting the extra- and intrahepatic bile ducts to all levels with multiple strictures. Adenomyosis of the gallbladder has also developed. AIH Score. To investigate whether differentiation between ASC and AIH on the basis of clinical, biochemical, and histological parameters, irrespective of cholangiographic changes, would have been possible, we have applied the scoring system proposed by the International Autoimmune Hepatitis Group. 21 Before immunosuppressive treatment, 23 (82%) patients with AIH, as well as 14 (52%) with ASC (P.017), fulfilled the criteria for the diagnosis of definite AIH (aggregate score 15); 5 patients with AIH and 13 with ASC were diagnosed as probable AIH. Of the 18 (13 ASC) patients diagnosed as probable AIH, 14 (10 ASC) were treated with immunosuppression, and 6 (4 ASC) would be reclassified as definite AIH (aggregate score of 17). In 3 patients (all ASC) who would have been classified pretreatment as definite AIH, 1 was not treated with immunosuppression, and 2 had no response to immunosuppression and were reclassified as probable AIH posttreatment. Overall, 25 (89%) patients with AIH, as well as 15 of 23 (65%) with ASC, would have been diagnosed as having definite AIH posttreatment (P.041). Patients With PSC. The clinical and laboratory features of the 9 children with PSC diagnosed during the observation period covered by this study are summarized in Table 6. Liver histology was diagnostic of sclerosing cholangitis in 5, biliary cirrhosis in 2, and chronic hepatitis in 1, while 1 child had nonspecific portal inflammation. Eight patients had an AIH score 9, while 1 scored 15 ( probable AIH). Four children were treated with UDCA; 1 with colchicine; 1 with colchicine and penicillamine; 2 did not receive treatment for liver disease, although 1 was treated with steroids and salazopyrine for associated inflammatory bowel disease; 1 child, on prednisolone and azathioprine at referral because of histological features of chronic hepatitis, continued with these drugs with addition of UDCA. Two children were lost to follow-up, the follow-up for the remaining 7 children being 5 to 15 years (median, 6 years). At the end of December 1999, 1 child had died of ischemic hepatitis 9 years after diagnosis; 6 children were alive, 4 with normal liver function tests on UDCA alone (2 with hepato-splenomegaly) and 2 with abnormal liver function tests. Of these 2 children, 1 is the child with chronic hepatitis on initial biopsy and an AIH score of 15, who was treated with immunosuppression and UDCA. Four years after presentation, she became LKM1-positive and had a steroidresponsive relapse. The other is a child on steroid treatment for inflammatory bowel disease, who required cholecystec-

7 550 GREGORIO ET AL. HEPATOLOGY March 2001 TABLE 6. Clinical and Laboratory Features of 9 Patients With PSC Age at diagnosis (yr), median (range) 6.6 (2-14.5) Male 6 Mode of presentation Prolonged acute hepatitis 1 No history of jaundice 8 Signs at diagnosis Hepatomegaly 7 Splenomegaly 6 Hepatosplenomegaly 6 Symptoms at diagnosis Diarrhea 6 (3 bloody) Abdominal pain 3 Fatigue 2 Pruritus 1 Weight loss 2 Associated autoimmune disorders 4 Ulcerative colitis 3 Autoimmune thyroiditis 1 Autoimmune disorders in first-degree relatives 3 Total bilirubin, mol/l (nv: 20) 15 (5-26) AST, IU/L (nv: 50) 90 (26-760) GGT, IU/L (nv: 50) 141 (23-688) AP, IU/L (nv: 350) 474 (23-688) AP/AST ratio 5.5 ( ) Total protein, g/l (nv: 60-80) 70 (52-73) Albumin, g/l (nv: 30-50) 40 (31-47) INR (nv: ) 1.0 ( ) IgG, g/l (nv: 5-18) 10.8 (5.7-23) IgA, g/l (nv: ) 1.12 ( ) IgM, g/l (nv: 0.5-2) 1.25 ( ) ANCA-positive (perinuclear pattern) 4 NOTE. Values expressed as median (range). Abbreviation: nv, normal value. tomy for mucocele of the gallbladder 8 years after presentation. DISCUSSION Cholangiography performed at presentation in children with clinical, biochemical, and histological manifestations typical of AIH has shown that half of them have bile duct abnormalities diagnostic of sclerosing cholangitis. The prospective nature of the present study has allowed us to document the appearance of bile duct changes typical of sclerosing cholangitis in 1 child who had a normal biliary tree at presentation 8 years earlier. This demonstrates the evolution from one condition to the other. An association between sclerosing cholangitis and AIH has been suggested previously. 4-8 Anecdotal cases of sclerosing cholangitis have been reported in which cholangiography was performed early in the course of the disease in patients with positive autoimmune serology, because of a mixed biliary and hepatitic histological picture. 5,8,24 Had this study followed the same indication to perform a cholangiogram, the diagnosis of sclerosing cholangitis would have been missed in over one third of our patients. Possible progression from AIH to sclerosing cholangitis has been suggested in a small number of adult patients, 5-7 but radiology was performed late in the investigative sequence, when follow-up liver biopsies, with or without biochemical signs of cholestasis, showed the appearance of biliary changes. The diagnosis of AIH is based on the presence of increased aminotransferase activity associated with high IgG, positive autoantibodies, and histological evidence of periportal hepatitis, in the absence of an infective or toxic cause of liver damage. 9 All children in this series fulfilled the majority of these criteria, including those with radiological evidence of cholangiopathy. At presentation, 52% of children with ASC satisfied internationally agreed criteria 21 for the diagnosis of definite and 48% of probable AIH, indicating that the AIH scoring system does not allow discrimination between AIH and ASC, unless there is visualization of the biliary tree. One of the aims of this study was to investigate whether clinical, biochemical, and histological features differ at presentation and follow-up according to the presence or absence of bile duct disease. A female preponderance was observed in both conditions, though it was less marked in ASC than in AIH. The mode of presentation was similar, but, surprisingly, pruritus was less common in ASC, and inflammatory bowel disease, though more frequent in ASC, was also present in 18% of children with ANA-/SMA-positive AIH. There was a considerable overlap in biochemical parameters between the 2 conditions at presentation, although impairment in liver synthetic function, as documented by increased INR and decreased serum albumin concentration, and evidence of hepatocellular dysfunction, as assessed by increased AST and bilirubin levels, were significantly more severe in AIH. GGT and AP levels, biochemical signs of cholestasis, were unexpectedly similarly elevated in the 2 conditions, while 7 children with ASC had both values within the normal range. Normal AP has been previously reported in 50% of children with sclerosing cholangitis of various etiologies, 3 and appears to be a feature peculiar to childhood, because normal AP occurs in only 3% of adult patients with PSC. 24,25 The only biochemical change significantly more frequent in ASC than in AIH in our series was a ratio exceeding 3 between absolute values of AP and AST. Analysis of the autoantibody profile showed that ANA, SMA, or a combination of the two are equally frequent, and their titers similar, in ASC and AIH, while only 1 of the 9 LKM1-positive patients had an abnormal cholangiogram at presentation. Seropositivity for panca, though more frequent in patients with sclerosing cholangitis, was also common in ANA-/SMA-positive AIH, akin to reports in adults. 22 Among the HLA alleles reported to confer susceptibility (DR3 and DR4) or resistance (DR15) to AIH 26 and susceptibility (DR3, DR13 and DR15) or resistance (DR4) to sclerosing cholangitis, 26 DR3 was increased significantly in AIH, but not in ASC, while in both conditions, DR4 was significantly decreased and frequencies of DR13 and DR15 were neither lower nor higher than controls. A decreased frequency of DR4 in juvenile AIH 9,27 and in adult PSC has been reported previously. 28 In the latter condition, however, possession of this allele has been linked to a more rapid disease progression, 29 an observation not confirmed in the present series or by others. 30,31 The main histological feature at presentation was interface hepatitis, which was similarly frequent in AIH and ASC. Also similar was the degree of fibrosis, while a biliary pattern of periportal hepatitis and cholangitic changes were more frequent in ASC, though not exclusive to this condition. Classical periductal onion-skin fibrosis was rare, being present only in 2 patients with ASC. Concordance between histological and radiological diagnosis was incomplete: 35% of the patients with ASC had histological features of chronic hepati-

8 HEPATOLOGY Vol. 33, No. 3, 2001 GREGORIO ET AL. 551 tis with no bile duct damage, and would have been diagnosed as AIH had we not performed an ERCP at onset. Conversely, 31% of the patients with AIH who had biliary changes on initial liver biopsy had normal cholangiographic studies. These patients, and in particular, the only girl with AIH diagnosed histologically as sclerosing cholangitis, may have small duct disease 32 or the autoimmune cholangitis syndrome recently described by Czaja et al., 33 in which autoantibody-positive adult patients with cholestatic clinical, laboratory, and/or histological changes had a normal biliary tree as assessed by ultrasound and/or cholangiography. In contrast to Czaja s patients, however, our AIH patients with histological biliary changes responded satisfactorily to immunosuppression. Interestingly, of 7 of these patients who underwent follow-up biopsies, only 3 continued to have biliary changes. One third of the patients with ASC had cholangiopathy limited to the intrahepatic system, akin to what has been previously reported in children with sclerosing cholangitis of various etiologies. 3 Of note, however, is that 2 of our patients with only intrahepatic lesions initially progressed to have intra- and extrahepatic bile duct lesions on follow-up cholangiograms, supporting the view that intrahepatic changes are the early manifestation of a process that may ultimately involve extrahepatic bile ducts. Though this study shows that ERCP is a safe procedure in expert hands, mild pancreatitis was observed in 7% of our patients. Magnetic resonance cholangiography would be a safer option, but while it would allow diagnosis of grade 3 and 4 changes, it would not visualize minor ductal abnormalities of the third- and fourth-order ducts. Ninety-five percent of our patients, including those with cholangiopathy, were treated with the immunosuppressive protocol used in our Unit for AIH. 9 This therapeutic approach was based on the similarities between the 2 conditions and on our previous observation of a favorable response to immunosuppression in children with sclerosing cholangitis and autoimmune features. 4 Response to this treatment was satisfactory and similar in the 2 conditions with respect to remission and relapse rates, times to normalization of biochemical parameters, and decreased inflammatory activity on follow-up liver biopsies. However, though most follow-up ERCPs were unchanged, none showed regression, while 8 children with ASC had progression of cholangiopathy and 1 child with AIH developed sclerosing cholangitis 8 years after presentation, suggesting that prednisolone and azathioprine are effective in abating the parenchymal inflammatory damage, but may not be as effective in controlling the bile duct disease. The positive response to immunosuppressive treatment observed in our patients with ASC contrasts with the disappointing results reported in a small number of children with sclerosing cholangitis associated with autoimmune features, 3 possibly because, at the time of diagnosis, these children had a more advanced disease than those recruited in our prospective study. Since 1992, following early reports of a beneficial effect in adult PSC, most of our patients with ASC received UDCA. Three children, treated with UDCA alone, showed no histological deterioration on follow-up liver biopsies, and 2 of them, who had follow-up ERCPs, showed no progression of bile duct damage. The small number of patients and the relatively short follow-up do not allow us to determine whether children with ASC treated with UDCA from onset have a better outcome than that recently reported in adult patients with established PSC treated with this drug. 34 Medium-term prognosis in our series is good, all patients being alive at a median follow-up of 7 years. Despite biochemical features of a more aggressive disease at presentation, none of the children with AIH needed transplantation, while 4 children with ASC required liver transplantation between 2 and 11 years from diagnosis, suggesting that the presence of cholangiopathy worsens the prognosis. Biochemical indices of liver function, including AST, AP, and GGT, appear to be inadequate prognostic markers for bile duct disease. It is therefore possible that some children with no follow-up ERCP may have progressive biliary damage. None of our patients has developed cholangiocarcinoma, possibly reflecting the relatively short follow-up of this study. However, in adults with PSC, this complication develops in 7% to 30% within 10 years from diagnosis. 35 It is possible that early therapeutic control of the inflammatory process prevents or delays the onset of cholangiocarcinoma. The only ASC patient in our series with bile duct dysplastic changes in the baseline liver biopsy, which in adults predict the development of cholangiocarcinoma, refused follow-up ERCP. Two questions remain to be answered: Are ASC and AIH different diseases or different manifestations of the same disease process? What is the relationship between ASC and PSC? In answer to the first question, our data suggest that ASC and AIH in childhood lie within the spectrum of the same disease process. The 2 conditions are very similar in their mode of presentation and response to treatment, the only clear difference between them being the involvement of the biliary tree in the disease process of ASC. It remains to be clarified why, in some patients with autoimmune liver disease, the immunopathological damage is confined to the liver parenchyma, while in others, the inflammatory process also involves the bile ducts. In patients with sclerosing cholangitis and inflammatory bowel disease, it has been suggested that the damage to the bile ducts derives from portal bacteremia or absorption of colonic toxins, in genetically predisposed individuals. 1 Notably, the girl in our series who progressed from AIH to ASC had ulcerative colitis from presentation. With respect to the second question, our data show that ASC is the most frequent cause of sclerosing cholangitis in childhood and is more common than PSC, only 9 patients having been diagnosed as having PSC during the observation period of this study. The converse is true in adults, in whom AIH/PSC overlap syndrome represents 6% to 8% of all PSC cases in recent reports. 36,37 In one of these studies, however, when the AIH scoring system was applied, a diagnosis of probable or definite AIH was made in 22% of 113 PSC patients. 37 It is conceivable that in adults, as in pediatric patients, sclerosing cholangitis is the consequence of different pathologies, and that at least some adult PSC cases are an advanced, at times burnt-out, stage of ASC. Features of AIH have been repeatedly reported in adult patients with PSC, ANA, and/or SMA being present in about 20% of patients, high levels of IgG in 60% of patients, and histological features of chronic hepatitis in 30% of patients. 38,39 In addition, panca is present in 50% to 90% of adults with either PSC or AIH. 22,40 One of our patients diagnosed as having PSC because of the absence of serological evidence of autoimmunity at referral developed LKM1 antibody 4 years later during a biochemical relapse, which responded to an increased dose of

9 552 GREGORIO ET AL. HEPATOLOGY March 2001 immunosuppression. It is conceivable that this patient, who had been treated with steroids and azathioprine in her local hospital before referral to our Unit, did in fact have ASC, but that the autoimmune features were masked by treatment. Differences between children with ASC and adults with PSC are the sex distribution and the response to immunosuppression. While in adults there is a male preponderance and the response to immunosuppressive treatment is reportedly poor, 39 in our series, 55% of the patients with ASC are female and the majority has had a favorable response to prednisolone and azathioprine. The different sex distribution is not readily explained: one possibility is that the disease is self-limiting in females and progressive in males. Because the evolution of PSC is quite slow the median reported survival in adults with this condition being 10 to 12 years from the time of diagnosis 10,41 a much longer follow-up of our patients is necessary to verify this hypothesis. Another possibility is that ASC and PSC are distinct entities, as suggested by the fact that a male preponderance was also observed in our patients with PSC who had no evidence of autoimmunity, though their number is too small to draw firm conclusions. Interestingly, in a recent publication, the majority of adults with AIH/PSC overlap syndrome were female, 36 akin to our findings, though this was not confirmed in another series. 37 The difference in response to immunosuppressive treatment between children with ASC and adults with PSC could be a result of the fact that adult PSC encompasses different etiologies, only some having an autoimmune component, or that sclerosing cholangitis diagnosed in adult life is too advanced to respond to treatment. A number of reports do show that adult patients with AIH/ PSC overlap syndrome can respond satisfactorily to immunosuppressive treatment, 37,38,42 although systematic studies using prednisolone and azathioprine in a large number of these patients have not been performed. In conclusion, our data show that ANA-/SMA-positive AIH and ASC are likely to belong to the same disease process, and that ASC and at least some cases of PSC largely overlap. We propose, however, the use of the term autoimmune sclerosing cholangitis as a reminder that this condition responds satisfactorily to immunosuppression. Acknowledgment: The authors thank Drs. Alex Gimson, Martin Lombard, John Ramage, David Westaby, Tony Ellis, John Devlin, and Mark Wilkinson for performing the ERCP studies, and Dr. Yun Ma for measuring CYP2D6 antibodies. REFERENCES 1. Angulo P, Lindor KD. Primary sclerosing cholangitis. HEPATOLOGY 1999; 30: Debray D, Pariente D, Urvoas E, Hadchouel M, Bernard O. Sclerosing cholangitis in children. J Pediatr 1994;124: Wilchanski M, Chait P, Wade JA, Davis L, Corey M, St Louis P, Griffiths A, et al. Primary sclerosing cholangitis in 32 children: clinical, laboratory and radiographic features with survival analysis. HEPATOLOGY 1995;22: El-Shabrawi M, Wilkinson ML, Portmann B, Mieli-Vergani G, Chong SKF, Williams R, Mowat AP, et al. Primary sclerosing cholangitis in childhood. Gastroenterology 1987;92: McNair ANB, Maloney M, Portmann BC, Williams R, McFarlane IG. Autoimmune hepatitis overlapping with primary sclerosing cholangitis in five cases. Am J Gastroenterol 1998;93: Gohlke F, Lohse AW, Dienes HP, Lohr H, Marker-Hermann E, Gerken G, Meyer zum Büschenfelde KH. Evidence for an overlap syndrome of autoimmune hepatitis and primary sclerosing cholangitis. J Hepatol 1996; 24: Lawrence SP, Sherman KE, Lawson M, Goodman ZD. A 39 year old with chronic hepatitis. Semin Liver Dis 1994;14: Rabinovitz M, Demetris AJ, Bou-Abboud CF, Van Thiel DH. Simultaneous occurrence of primary sclerosing cholangitis and autoimmune chronic active hepatitis in a patient with ulcerative colitis. Dig Dis Sci 1992;37: Gregorio GV, Portmann B, Reid F, Donaldson PT, Doherty DG, McCartney M, Mowat AP, et al. Autoimmune hepatitis in childhood: a 20-year experience. HEPATOLOGY 1997;25: Farrant JM, Hayllar KM, Wilkinson ML, Karani J, Portmann B, Westaby D, Williams R. Natural history and prognostic variables in primary sclerosing cholangitis. Gastroenterology 1991;100: Desmet VJ, Gerber M, Hoofnagle JH, Manns M, Scheuer PJ. Classification of chronic hepatitis: diagnosis, grading and staging. HEPATOLOGY 1994; 19: Portmann B, Popper H, Neuberger J, Williams R. Sequential and diagnostic features in primary biliary cirrhosis based on serial histologic study in 209 patients. Gastroenterology 1985;88: Holborow J, Johnson G. Antinuclear factor in systemic lupus erythematosus. A consideration of the immunofluorescent method of detecting antinuclear antibodies, with results obtained in a family study. Arthritis Rheum 1964;7: Ma Y, Gregorio GV, Gäken J, Muratori L, Bianchi FB, Mieli-Vergani G, Vergani D. Establishment of a novel radioligand assay using eukaryotically expressed cytochrome P4502D6 for the measurement of liver kidney microsomal type 1 antibody in patients with autoimmune hepatitis and hepatitis C virus infection. J Hepatol 1997;26: Terasaki PI, McClelland JD, Park MS, McCurdy B. Microdroplet assay of human serum cytotoxins. In: Ray JG, Hare DB, Pederson PD, Kayhoe DE, eds. Manual of Tissue Typing Techniques. Washington, DC: DHEW Publications, 1974: Donaldson PT, Doherty DG, Haullar KM, McFarlane IG, Johnson PJ, Williams R. Susceptibility to autoimmune chronic active hepatitis: human leukocyte antigens DR4 and A1-B8-DR3 are independent risk factors. HEPATOLOGY 1991;13: Hayashi H, Higuchi T, Ichimiya H, Hishida N, Sakamoto N. Asymptomatic primary sclerosing cholangitis treated with ursodeoxycholic acid. Gastroenterology 1990;99: Beuers U, Spengler U, Kruis W, Aydemir U, Wiebecke B, Heldwein W, Weinzierl M, et al. Ursodeoxycholic acid for treatment of primary sclerosing cholangitis: a placebo-controlled trial. HEPATOLOGY 1992;16: Lebovics E, Salama M, Elhosseiny A, Rosenthal WS. Resolution of radiographic abnormalities with ursodeoxycholic acid therapy of primary sclerosing cholangitis. Gastroenterology 1992;102: Woolf B. On estimating the relation between blood group and disease. Ann Hum Genet 1955;19: International Autoimmune Hepatitis Group Report: review of criteria for diagnosis of autoimmune hepatitis. J Hepatol 1999:31; Angulo P, Peter JB, Gershwin ME, DeSotel CK, Shoenfeld Y, Ahmed AEE, Lindor K. Serum autoantibodies in patients with primary sclerosing cholangitis. J Hepatol 2000:32; Hartshorne N, Hartman G, Markin RS, Demetris AJ, Ferrell L. Bile duct hemorrhage: a biopsy finding after cholangiography or biliary tree manipulation. Liver 1992;12: Balasubramanian K, Wiesner RH, LaRusso NF. Primary sclerosing cholangitis with normal serum alkaline phosphatase activity. Gastroenterology 1988;95: Cooper JF, Brand EJ. Symptomatic sclerosing cholangitis in patients with a normal alkaline phosphatase: two case reports and a review of the literature. Am J Gastroenterol 1988;83: Donaldson PT, Manns MP. Immunogenetics of liver disease. In: Bircher J, Benhamou J-P, McIntyre N, Rizzetto M, Rhodes J, eds. Oxford Textbook of Clinical Hepatology. Oxford: Oxford University Press, 1999: Doherty DG, Donaldson PT, Underhill JA, Farrant JM, Duthie A, Mieli- Vergani G, McFarlane IG, et al. Allelic sequence variation in the HLA class II genes and proteins in patients with autoimmune hepatitis. HEPA- TOLOGY 1994;19: Farrant JM, Doherty DG, Donaldson PT, Vaughan RW, Hayllar KM, Welsh KI, Eddleston ALWF, et al. Amino acid substitutions at position 38 of the DRb polypeptide confers susceptibility and protection from primary scelroding cholangitis. HEPATOLOGY 1992;16: Mehal WZ, Lo DY-M, Wordworth P, Lo, YMD, Wordsworth BP, Neuberger JM, Hubscher SC, et al. HLA DR4 is a marker for rapid disease progression in primary sclerosing cholangitis. Gastroenterology 1994; 106: