Prevalence of sclerosing cholangitis in adults with autoimmune hepatitis: Evaluating the role of routine magnetic resonance imaging


  • Potential conflict of interest: Dr. Heathcote is a consultant for, advises, is on the speakers bureau of, and received grants from Axcan-Pharma.


Large bile duct injury (that seen on cholangiography) is not usually considered a feature of autoimmune hepatitis (AIH) in adults but is present in up to 50% of children with AIH. The aim of this work was to study the prevalence of large bile duct abnormalities identified by magnetic resonance cholangiography (MRC) in adults given a diagnosis of AIH. Seventy-nine (n = 79) patients given a diagnosis of AIH (mean AIH score: 15.1 ± 3.4) were screened with MRC for evidence of sclerosing cholangitis (SC). Results were reviewed by two radiologists. Clinical parameters were correlated with MRC findings. A histological review of available liver biopsies (n = 29) was performed. Of the 79 patients surveyed, 8 (10%) had MRC findings consistent with primary sclerosing cholangitis (PSC). The interrater variability was excellent (kappa = 0.87). Younger age at diagnosis (24.3 ± 11.9), higher baseline alkaline phosphatase (186.4 ± 98.3), higher bilirubin at time of MRC (45.8 ± 37.2), and greater lobular activity on initial liver biopsy were significantly associated with the detection of this overlap of SC with AIH (P = 0.024, P = 0.037, P = 0.032, and P = 0.041, respectively), but not alkaline phosphatase/aspartate aminotransferase ratio, time between the initial diagnosis of AIH and the MRC, or the presence of cirrhosis on initial liver histology. Two cases with a normal MRC had histological lesions typical of small duct PSC. Conclusion: The presence of SC detected by MRC and from liver histology in adult patients with AIH may not be clinically overt, and thus the prevalence of this AIH/SC overlap may be higher than previously recognized. Our data suggest that routine radiological evaluation of the biliary tree should be performed in adults given a diagnosis of AIH, as in children the presence of this overlap negatively impacts on survival. (HEPATOLOGY 2008.)

Autoimmune liver disease encompasses a heterogeneous spectrum of disorders with certain specific characteristics yet sometimes with overlapping clinical, biochemical, serologic, histologic, and radiographic characteristics. Autoimmune hepatitis (AIH), primary biliary cirrhosis (PBC), and primary sclerosing cholangitis (PSC) are the predominant presumed autoimmune liver diseases. Although these diseases may be easily distinguishable on clinical grounds, occasionally patients will exhibit features suggestive of an overlap syndrome. This latter term is particularly valid when two autoimmune liver diseases are either present simultaneously or consecutively. The International Autoimmune Hepatitis Group developed an AIH score in 1993, revised in 1999, that included descriptive criteria for definite and probable diagnoses of AIH based on a numeric scoring system,1, 2 but there is no single gold standard for AIH. The diagnosis of PBC is based predominantly on laboratory profile: antimitochondrial antibodies (AMAs) and elevated alkaline phosphatase (ALP). The diagnosis of large duct PSC depends entirely on the cholangiographic demonstration of multifocal strictures of the hepatobiliary tree.3

Wilschanski et al.4 reported 32 children who were initially given a diagnosis of PSC based on endoscopic retrograde cholangiopancreatography (ERCP) findings; 9 were noted to have histologic features of AIH. Subsequently, Gregorio et al.5 made a prospective study of a large series of children with clinically overt AIH and found that 27 of 55 (49%), all of whom underwent ERCP at the time of initial diagnosis of AIH, had cholangiographic evidence of what the authors termed autoimmune sclerosing cholangitis. The only case series in adults was reported by Abdo et al.,6 who described 6 adults initially given a diagnosis of AIH, 3 of whom had undergone ERCP at initial presentation. All 6 had no histological evidence of duct involvement on their baseline liver biopsy. Subsequent failure to respond to immunosuppression prompted either repeat or first-time ERCP, which revealed typical cholangiographic evidence of PSC in all 6.

ERCP has long been considered the gold standard for the diagnosis of PSC. Although it provides high-quality images of the biliary tract, the procedure can be associated with significant risks.7, 8 Alternatively, magnetic resonance cholangiography (MRC) is a noninvasive, rapidly performed imaging modality that may be safely used in the evaluation of the biliary tree. Fulcher et al.9 recently reported that the sensitivity and specificity of MRC in the detection and localization of PSC were as high as 88% and 97%, respectively.9

This study reports the results of a routine evaluation of the biliary tree via MRC in patients with a diagnosis of AIH. We have examined the prevalence of large bile duct abnormalities in adults given a diagnosis of AIH and have determined demographic, clinical, and biochemical features associated with overlapping AIH and sclerosing cholangitis (SC).


AIH, autoimmune hepatitis; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA, antimitochondrial antibody; ANA, antinuclear antibody; ASMA, anti–smooth muscle antibody; AST, aspartate aminotransferase; ERCP, endoscopic retrograde cholangiopancreatography; GGT, gamma glutamyl transpeptidase; IgG, immunoglobulin G; IgM, immunoglobulin M; INR, international normalized ratio; MR, magnetic resonance; MRC, magnetic resonance cholangiography; MRI, magnetic resonance imaging; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; SC, sclerosing cholangitis; SD, standard deviation; SMA, smooth muscle antibody; UDCA, ursodeoxycholic acid; US, ultrasonography.

Patients and Methods

Patients Studied.

Two-hundred four (n = 204) patients have attended our institution's AIH clinic between 1980 and 2005 at least once and have been given a diagnosis of AIH based on their clinical, biochemical, and histological profile. Seventy-nine (39%) patients have undergone MRC since we introduced routine evaluation of the biliary tree in 2004.

Following approval by our research ethics board, the medical charts of the patients with autoimmune liver disease were reviewed by two of the authors who were uninvolved in the interpretation of magnetic resonance imaging (MRI). A database was constructed that included age, gender, year of diagnosis, personal or family history of related autoimmune disorder(s), and biochemistry at diagnosis of AIH and at time of MRC [aminotransferases, ALP to aspartate aminotransferase (AST) ratio, albumin, bilirubin, international normalized ratio, immunoglobulin quantification, and autoantibody profile, that is, titers of antinuclear antibodies (ANAs), anti–smooth muscle antibodies, and AMAs].

Radiological Assessment.

MRC was performed with a 1.5T MRI system (Excite or Echospeed LX, GE Medical Systems, Milwaukee, WI) with a 4-element torso phased-array receiver coil. Both thick-slab and thin-slice MRC images were obtained. Thick-slab MRC was obtained by performing heavily T2-weighted breath hold single-shot fast-spin echo images obtained using a radial slab technique with a slice thickness of 4 to 6 cm, centered on the common hepatic duct in the porta hepatis. Sixteen slabs were acquired with a 22.5° radial interval to cover 360°. Each slab was acquired at end expiration with an interval of 8 seconds between each acquisition, a matrix of 320 × 320, an echo time (TE) of 990 milliseconds, a bandwidth of 62.5 kHz, a field of view of 28-32 cm, a phase/field of view of 1.0, and fat saturation. Bile is bright on these images because of its high water content, whereas the liver is very dark; thus, this technique yields a series of oblique coronal and sagittal projections of the bile ducts. Axial T2-weighted single-shot fast spin-echo images were also acquired with a TE of 180 milliseconds, a slice thickness of 4 mm, a bandwidth of 62.5 kHz, and a field of view of 28-32 cm.

MRC images were reviewed and interpreted independently by two radiologists whose area of expertise is the abdomen and who were blinded to patients' clinical profiles, including all historical, laboratory, pathological, and prior imaging findings. The magnetic resonance (MR) cholangiopancreatograms were independently interpreted for overall image quality, extent of ductal visualization, and the presence and location of bile duct strictures. The source images and the reconstructed three-dimensional views were assessed at an interactive workstation. The MR cholangiographic findings were recorded on standardized sheets. The certainty of the radiologist's interpretation was outlined through a three-point categorization system of definite, probable, or absent PSC (Fig. 1). Definite SC was defined by the presence of multisegment intrahepatic duct strictures in the absence of apparent cholangiocarcinoma. Probable SC implied the presence of intrahepatic duct irregularities in multiple segments without overt duct cutoff or obstruction. Criteria for a normal study included lack of apparent strictures and central and/or peripheral bile duct dilatation. Radiological presence of cirrhosis was made through assessment of regional changes in hepatic morphology.10 The presence of mild peripheral segmental biliary changes in areas of significant hepatic atrophy was recorded. The MRC pictures, obtained through the same imager, of an additional 13 patients with PBC were used as radiological controls. Following a review of the images obtained, the individual interpretations regarding the presence or absence of definite or probable PSC were compared to the patient's primary diagnosis.

Figure 1.

MR cholangiographic representative examples of “definite” PSC (A), probable PSC (B), and a normal MRC (C). (A) Coronal, thick-section RARE MR cholangiogram of the biliary tract reveals classic multisegment intrahepatic duct strictures (arrows) in a patient with “definite” PSC. (B) Coronal-oblique, thin-section half-Fourier RARE MR cholangiogram in a patient with “probable” PSC reveals multiple intrahepatic duct wall irregularities without obvious cutoff or obstruction. (C) Coronal-oblique, thin-section half-Fourier RARE MR cholangiogram in a patient with normal MRC.

Histological Assessment.

A single hepatopathologist reviewed the liver tissue specimens in batch under code and specifically evaluated the bile duct morphology according to criteria outlined by Czaja and Carpenter.11 The single-observer reproducibility of grading inflammatory indices on a single liver needle biopsy has been demonstrated previously.12, 13 Liver biopsy specimens obtained when the initial diagnosis of AIH was made were available in 29 (38%) patients. Fibrosis was staged according to the Laennec histological criteria.14 The Ishak grading system13 was used to assess hepatitic disease activity. Bile ducts were evaluated on routine trichrome, hematoxylin & eosin, and CK7 stains. The absence of a bile duct near an arteriole within the same portal tract indicated ductopenia. Ducts were individually counted, and the cutoff for duct loss was set at 10%.

Statistical Analysis.

The clinical parameters retrieved from our chart review were compared between patients with normal MRC versus abnormal MRC through a series of independent sample t tests and chi-square tests. The Fischer exact probability test was used to compare dichotomous variables, and the unpaired t test was used to compare differences in the means of continuous variables. Continuous measures were expressed as the mean ± standard deviation, whereas categorical features were summarized as n (%). Nonparametric variables in independent samples were compared by the Mann-Whitney test. To assess the degree of agreement between SC classifications provided by the two observers, a kappa statistic was calculated. The interpretation of this statistic was performed according to the guidelines outlined by Landis and Koch.15


Demographics and Clinical Features.

The main demographic and clinical data are outlined in Table 1. The patients included in our study were those who attended the clinic between 2004 and 2005. Sixty-three (79.8%) were women, and the mean age at presentation was 37.3 ± 17.6 years (range: 19-77 years). The time interval between when the patient was given a diagnosis of AIH and completion of the MRC ranged from 20 months to 15 years (mean: 8.4 years). Fifteen patients (19%) had concurrent illness of presumed autoimmune nature, including 2 patients with systemic lupus erythematosus, 3 patients with rheumatoid arthritis, 3 patients with Raynaud's disease, 2 patients with vitiligo, and 1 patient each with Sjogren's, scleroderma, and ulcerative colitis. Eight had thyroid dysfunction, one of whom had confirmed Grave's disease. In addition, 1 patient had autoimmune hemolytic anemia. These associations were not mutually exclusive, as several patients exhibited more than one autoimmune manifestation.

Table 1. Baseline Demographics: Patients with AIH Seen Between 2004 and 2005 (N = 79)
  • Abbreviations: AIH, autoimmune hepatitis; MRC, magnetic resonance cholangiography; SD, standard deviation.

  • *

    Based on revised International Autoimmune Hepatitis Group Criteria.2

Age at AIH diagnosis (years) (mean ± SD)37.3 ± 17.6
AIH score at diagnosis (mean ± SD)*15.1 ± 3.4
GenderFemale: 63 (79.8%)
Other autoimmune manifestations15 (19.0%)
Time interval between diagnosis and MRC20 months to 15 years (mean: 8.4 years)

All patients were graded by the revised scoring system proposed by the International Autoimmune Hepatitis Group,2 and only pretreatment scores at the time of their initial diagnosis were analyzed. A definite diagnosis of AIH by these criteria requires a pretreatment score > 15, and a probable diagnosis requires a score of 10 to 15 (inclusive). Our AIH score assessment did not incorporate the optional human leukocyte antigen DR3 or DR4 parameter or antineutrophil cytoplasmic antibody titers. Thirty-nine (49%) patients satisfied criteria for a definite diagnosis (pretreatment score: 19.1 ± 0.3, range: 16-22), and 37 patients (47%) satisfied scoring criteria for a probable diagnosis (pretreatment score: 13.2 ± 0.5, range: 10-15). The AIH score in those with SC on MRC did not differ from those with no cholangiographic evidence of SC (15.1 ± 3.5 versus 14.8 ±3.2, respectively, P = 0.712). Two patients did not meet scoring criteria for AIH. However, they would have scored at least as probable AIH had their tally not been downgraded by their AMA-positive status. One patient who was given a diagnosis of AIH in 1986 lacked appropriate documentation of her original liver biochemistry and antibody profile, which prevented us from tabulating her pretreatment score.

Table 2 outlines the biochemical findings of our entire study population both at presentation with AIH and at time of their MRC (when most patients had well-controlled liver disease). At baseline, serum levels of immunoglobulin G and immunoglobulin M were higher than normal in 81% and 22%, respectively. Thirty-nine (49.4%) had ANAs, 32 patients (41%) had smooth muscle antibodies (SMAs), and 25 patients (32%) had combined ANA and SMA positivity at presentation. Six patients (8%) tested AMA-positive at presentation, although their clinical profile, response to treatment, and serial liver histologies were more compatible with AIH and not PBC. The clinical and histological findings justified the diagnosis of AIH in each instance.

Table 2. Baseline Biochemistry: Entire Sample (N = 79)
Variable (Units; Normal Range)At AIH Diagnosis*At Time of MRC
  • Abbreviations: AIH, autoimmune hepatitis; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA, antimitochondrial antibody; ANA, antinuclear antibody; ASMA, anti–smooth muscle antibody; AST, aspartate aminotransferase; GGT, gamma glutamyl transpeptidase; IgG, immunoglobulin G; IgM, immunoglobulin M; INR, international normalized ratio; MRC, magnetic resonance cholangiography; N/A, not applicable; SD, standard deviation.

  • *

    The results are presented as mean ± SD and median (range) or proportion.

  • The results are presented as mean ± SD and median (range).

ALT (U/L; <40 U/L)324 ± 35369.4 ± 77.6
 195 (11, 1520)39 (11, 450)
ALP (U/L; <110 U/L)160 ± 124122 ± 104
 121 (45, 731)87 (32, 527)
AST (U/L; <35 U/L)321 ± 38374 ± 93
 152 (21, 1525)40 (18, 572)
ALP/AST1.2 ± 1.12.4 ± 1.8
 0.9 (0.1, 5.8)2.0 (0.2, 10.3)
GGT (U/L; <40 U/L)163 ± 138120 ± 139
 122 (7.0, 562)68 (0, 761)
Albumin (g/L; 38-50 g/L)38 ± 7.338 ± 5
 39 (19, 51)39 (26, 46)
INR (0.8-1.2)1.2 ± 0.31.1 ± 0.2
 1.1 (0.9, 2.5)1.1 (0.9, 1.7)
Bilirubin (μmol/L; <22 μmol/L)66 ± 9425 ± 46
 27 (2, 485)11 (5, 265)
IgG elevation (>13.3 g/L)64 (81.0%)N/A
IgM elevation (>2.2 g/L)17 (22.1%)N/A
ANA positivity39 (49.4%)N/A
ASMA positivity32 (42.1%)N/A
AMA positivity6 (8.3%)N/A

MRC Findings.

Of the 79 patients who underwent MRC, all but one were deemed to have scans of adequate quality. Ten of 79 patients (12.7%) were identified to have cholangiographic evidence of definite or probable SC (Table 3). Disagreement between the radiologists occurred in two cases, one of whom possessed the only poor quality scan. Both of these cases were omitted from the subgroup statistical analysis. The unweighted multirater kappa coefficient15 was 0.87, which corresponds to nearly perfect agreement between the two radiologists. Thus, the final number of patients with AIH who were shown to have overlapping SC on routine MRC was 8 (10.1%), 5 of whom were classified as having definite SC, whereas 3 were labeled as having probable disease. An additional 10 patients (half of whom had prior radiographic and histological evidence of cirrhosis) were reported to have mild peripheral segmental biliary changes in areas of hepatic atrophy; these findings were deemed nondiagnostic for SC by both radiologists, and similar changes were noted in 3 of 13 MRC images from patients with cirrhosis due to PBC. Mild irregularities of intrahepatic bile ducts that potentially mimic intrahepatic ductal changes suggestive of SC in patients with cirrhosis have been described previously by Fulcher et al.9

Table 3. MRC Results (N = 79)
  • Abbreviations: AIH, autoimmune hepatitis; MRC, magnetic resonance cholangiography; and SC, sclerosing cholangitis.

  • *

    Landis and Koch.15

Patients with AIH who underwent MRC79
Number of patients with SC identified on MRC10
Number of cases with interrater disagreement 2
Final number of AIH/SC overlap on MRC8 (5 definite and 3 probable)
Unweighted multirater kappa coefficient*0.87

Table 4 indicates the clinical and biochemical correlates found to be associated with large duct anomalies suggestive of SC on MRC. A younger age at AIH diagnosis (mean: 24.3 years), a higher baseline ALP (mean: 196.4 U/L), and a higher bilirubin at the time of MRC (mean: 45.8 μmol/L), despite patients appearing to be in remission of their AIH, were the only variables that reached statistical significance (P = 0.024, P = 0.037, and P = 0.032, respectively). The ALP at time of MRC, although elevated 2-fold (mean: 200.7 U/L), was not significantly associated with abnormal bile ducts on MRC. At presentation, 5 of the 8 patients (63%) who were shown to have SC on MRC had been classified as having definite AIH, whereas the other 3 patients (38%) had fulfilled criteria for probable disease (Table 5). Four patients had previously undergone ERCP primarily for cholestasis with or without symptoms, the results of which had suggested what was designated at that time as possible AIH/PSC overlap, prompting initiation of ursodeoxycholic acid (UDCA).

Table 4. Features of AIH Correlating with MRC Findings
Variable (Units; Normal Value)SC on MRC (n = 8; Mean ± SD)Normal MRC (n = 69; Mean ± SD)P Value
  • Abbreivations: AIH, autoimmune hepatitis; ALP, alkaline phosphatase; AST, aspartate aminotransferase; INR, international normalized ratio; MRC, magnetic resonance cholangiography; SC, sclerosing cholangitis; SD, standard deviation.

  • *

    Statistically significant.

Age at AIH diagnosis (years)24.3 ± 11.939.2 ± 17.70.024*
ALP at diagnosis of AIH (U/L; <110 U/L)196.4 ± 98.3150.3 ± 125.90.037*
ALP at MRC (U/L; <110 U/L)200.7 ± 162.8111.3 ± 93.80.052
Bilirubin at MRC (μmol/L; <22 μmol/L)45.8 ± 37.222.5 ± 47.90.032*
AST at MRC (U/L; <35 U/L)104.3 ± 74.271.3 ± 97.20.068
INR at MRC1.2 ± 0.21.1 ± 0.20.051
Cirrhosis on initial biopsy4 (50.0%)33 (48.5%)1.000
Table 5. Initial Clinical and Laboratory Features of AIH Patients with Noted SC on MRC (N = 8)
Parameter (Units; Normal Value)Value [Median (Range)]
  1. Abbreviations: AIH autoimmune hepatitis; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AMA, antimitochondrial antibody; ANA, antinuclear antibody; AST, aspartate aminotransferase; GGT, gamma glutamyl transpeptidase; IgG, immunoglobulin G; IgM, immunoglobulin M; INR, international normalized ratio; MRC, magnetic resonance cholangiography; PSC, primary sclerosing cholangitis; SC, sclerosing cholangitis; SMA, smooth muscle antibody; US, ultrasonography.

Age at diagnosis (years)19.5 (14–46)
Associated autoimmune disorders0
Ulcerative colitis1
Total bilirubin (μmol/L; <22 μmol/L)52.0 (11–142)
ALT (U/L; <40 U/L)276.0 (72–1205)
AST (U/L; <35 U/L)191.0 (60–890)
ALP (U/L; <110 U/L)187.5 (94–409)
GGT (U/L; <40 U/L)234.0 (47–503)
ALP/AST ratio0.7 (0.2–3.9)
Albumin (g/L; 38–50 g/L)39.0 (19–43)
INR (0.8–1.2)1.2 (1.0–2.5)
Elevated IgG (>13.3 g/L)7
Elevated IgM (>2.2 g/L)2
ANA positivity5
SMA positivity3
AMA positivity0
Cirrhosis on biopsy4
Cirrhosis on US5
AIH score15.0 (10–19)
Definite PSC (number of patients)5
Probable PSC (number of patients)3
Treatment response5
Mean interval between AIH diagnosis and MRC7.4 years

Histological Review.

Of the original liver biopsies that were available to be reassessed (n = 29), 3 belonged to patients who were subsequently identified as having definite SC on MRC, whereas the remaining 26 were of those whose MRC suggested no SC. The incidence of cirrhosis and the presence of destructive/nondestructive cholangitis as described by Czaja and Carpenter11 did not differ between the two groups (Table 6). However, even though the numbers are small, patients with SC on MRC were more likely to initially demonstrate a higher degree of lobular activity (P = 0.041). Two patients with a normal MRC had distinct bile duct changes suggestive of the concentric periductal onion-skin fibrosis seen in PSC (Fig. 2), raising the possibility of small duct PSC in these two additional cases.

Table 6. Histological Findings
HistologyPSC (n = 3)Normal (n = 26)P Value
  • Abbreviations: N/A not applicable; PSC, primary sclerosing cholangitis; SD, standard deviation.

  • *

    Statistically significant.

Cirrhosis2 (66.7%)19 (73.4%)0.423
Lobular activity score1:0 (0.0%)1:12 (46.2%)0.041*
 2:0 (0.0%)2:6 (23.1%) 
 3:2 (66.7%)3:7 (26.9%) 
 4:1 (33.3%)4:1 (3.9%) 
Destructive/nondestructive cholangitis1 (33.3%)8 (30.8%)1.000
Duct loss (mean ± SD)8.3% ± 7.2%11.0% ± 9.5%N/A
PSC changes0 (0.00%)2 (7.7%)1.000
Figure 2.

Trichrome stain of the original liver biopsies in 2 AIH patients with subsequent normal MRC. In both, large septal ducts are surrounded by extensive collageneous connective tissue elements with minimal inflammation, generating a pattern of concentric periportal fibrosis (arrows) in an “onion skin” morphology suggestive of PSC.


MRC performed an average of 8.4 years after an initial diagnosis of AIH revealed large bile duct abnormalities diagnostic of SC in 8 of 79 consecutive adult patients surveyed (10%) and small duct changes in another 2 patients. Heightened diagnostic awareness in recent years has suggested an association between AIH and SC.18–25 Most published reports of the variant form of AIH with PSC overlap describe patients in whom cholangiography was performed because of mixed hepatitic and cholestatic biochemical and/or histological findings. Had our study followed a similar indication to perform a cholangiogram, the diagnosis of SC would have been missed in half of the patients subsequently identified as having SC on MRC.

The frequency of possible AIH in patients given a primary diagnosis of PSC has been indirectly assessed retrospectively. Czaja16 reported a positive AIH score in 54% of 26 patients with PSC surveyed, and Boberg et al.17 also conducted a post hoc AIH score determination of 114 adults with PSC: 38 had probable scores and 2 had definite scores for an AIH diagnosis. However, when the revised international scoring system was later employed, only 10 of the 38 retained a score of probable AIH. Later, Gregorio et al.'s5 prospective study employed ERCP at the time of initial diagnosis of 55 children with AIH and revealed an AIH-SC overlap in close to 50% of those surveyed. In adults, systematic investigations of AIH-SC overlap are scarce. Abdo et al.6 described 6 adult patients diagnosed with SC many years after well-established AIH. This overlap was identified only when severe cholestasis and/or resistance to standard immunosuppressive therapy was observed a mean of 4.6 years following the initial diagnosis and initiation of treatment for AIH. The authors could not identify any specific features at baseline that predicted this subsequent overlap. A recent study compared the clinical features of 7 patients with AIH-PSC with 34 patients with classical PSC.26 Patients with the overlap syndrome were younger at presentation (mean: age 21 versus 32), had significantly higher serum aminotransferase and immunoglobulin levels, and were diagnosed with SC only when immunosuppression had failed to achieve clinical remission of AIH. In our study, a younger age at diagnosis of AIH, a higher baseline ALP, and a higher bilirubin at the time of imaging were significantly associated with the finding of intrahepatic large duct disease on MRC. Half (4/8) were in complete biochemical remission at the time their MRC-indicated SC. Of note, the ALP/AST ratio, previously demonstrated to be of clinical significance in children,5 was not of diagnostic value in our adult population.

International criteria for the diagnosis of AIH do not recognize histological features of bile duct injury as part of the syndrome, and yet a recent study has indicated that patients with AIH may be found to have biliary changes on liver biopsy.11 Christofferson et al.27 first described bile duct changes in chronic hepatitis, but their initial investigations were performed before the development of concise diagnostic classifications and adequate screening methods to rule out biliary tract disease. Czaja and Carpenter11 systematically evaluated the histological features of patients satisfying international criteria for AIH to determine the frequency and clinical significance of destructive and nondestructive bile duct changes in classical AIH. Twenty patients (24%) were noted to have biliary changes, including 6 with destructive cholangitis, 4 with ductopenia, and 10 with nondestructive cholangitis. Patients with or without bile duct changes had similar biochemical tests. The authors postulated that the simultaneous occurrence of bile duct changes with features of AIH might represent a form of AIH with evolving PBC or PSC, concurrent AIH and PBC or PSC, or AIH with incidental collateral bile duct injury. Czaja and Carpenter's analysis also demonstrated that adult patients with AIH who had incidental small duct changes did not differ in overall treatment outcome and steroid responsiveness in comparison to the group lacking such features. None of their 24 patients with incidental small duct changes tested AMA-positive. Only 6 subsequently underwent ERCP, none of whom were found to have cholangiographic evidence of large duct abnormalities. None were reported to have the onion-skin bile ducts, as was observed in 2 of our patients thought to have small duct PSC in addition to their AIH.

ERCP has long been considered the gold standard for the diagnosis of SC. Ultrasonography and computed tomography may indicate thickening of the bile duct wall and biliary dilatation,28, 29 but these modalities are not recommended for the diagnosis of SC. ERCP provides high-quality images of the biliary tract, yet this invasive procedure is associated with risks of sepsis and pancreatitis.30 Advantages of ERCP include a high sensitivity to detect intrahepatic peripheral duct abnormalities and the provision of interventional therapeutic capabilities. MRC is a rapidly performed noninvasive examination that has no attendant complications, provided that the patient is prescreened for contraindications. Data from 10 years ago show that it is comparable to ERCP in the detection and localization of bile duct obstruction and choledocholithiasis.31–33 By the use of heavily T2-weighted sequences, the signal of static or slow-moving fluid-filled structures such as the bile and pancreatic ducts has been greatly increased, and this has resulted in a heightened duct-to-background contrast.31, 34 MRC is a much more suitable screening tool than ERCP as it is a noninvasive and hence safer and less expensive modality than endoscopic intervention. It employs nonionizing radiation and does not require the use of sedative drugs. It has the great advantage of highlighting ducts proximal to an obstruction or tight stenosis better than conventional cholangiography.35 However, MRC provides less spatial resolution, leading to higher thresholds of detection of peripheral duct abnormalities. Smaller ducts are also not adequately visualized because imaging is performed in the physiologic nondistended state. The predominant caveat to MRC imaging is the delay that it incurs in patients who need a therapeutic intervention.

The role of MRC in the evaluation of SC and intrahepatic duct abnormalities has been presented. Two reports suggest a good correlation between MRC and ERCP in the diagnosis of SC.31, 36 In a prospective case-control study involving 102 patients, Fulcher et al.9 compared the MR cholangiograms obtained in 34 patients with PSC to those of 68 age-matched control patients with various hepatobiliary diseases, 6 of whom had AIH. In the detection of intrahepatic PSC, the reported sensitivity and specificity of MRC were between 85% and 88% and between 92% and 97%, respectively. Positive predictive values ranged between 85% and 94%. There was a reported high rate of visualization of the first-order and second-order bile ducts (94% and 97%, respectively, in the PSC group), whereas the third-order and fourth-order ducts were seen less frequently (85% and 53%, respectively, in the group with PSC). Partial depiction of the extrahepatic biliary system was achieved in patients with PSC, and complete depiction was seen in all the control patients (74% versus 100% of patients, respectively). Of the 5 patients in our study with false positive features of SC on MRC, all had clinical, pathologic, and/or imaging evidence of background cirrhosis that resulted in distortion of the intrahepatic ducts seen on MRC. Our MRC controls were patients with PBC, 3 of whom had similar peripheral segmental biliary changes in areas of hepatic atrophy.

The 10-year survival of treated adults with AIH is excellent.37 The survival of adults with AIH and overlapping PSC has been compared to that of PSC, not AIH without PSC. In the study by Gregorio et al.,5 the presence of cholangiopathy was associated with a worse prognosis over the 16 years of follow-up; 4 of the 27 children with SC required liver transplant, whereas none of the 28 who lacked evidence of large duct biliary changes did. The survival analysis by Feldstein et al.38 of children with PSC revealed a survival disadvantage over an age-matched and gender-matched US population, regardless of the presence or absence of concomitant AIH. Pharmacological therapy did not seem to impact survival and long-term outcome in their longitudinal cohort assessment of 16 years. In adults, survival and treatment data are equally scarce. In a report from the Mayo Clinic that included 9 adult patients with PSC and a positive AIH score, clinical remission was observed in only 2 patients following steroid therapy.16 Response was more likely if patients had lower levels of ALP and higher serum concentrations of total gamma globulin and immunoglobulin G. In a more recent study26 in which the clinical course of 7 adult patients with AIH-SC was compared with that of 34 patients with classical PSC, a more favorable outcome was observed in those with the overlap syndrome, but no comparison was made with the survival of those patients with AIH alone.26 Treatment of AIH/SC overlap in this series reported by Floreani et al.26 included a combination of UDCA, prednisone, and azathioprine, whereas patients with classical PSC were administered UDCA alone. The Mayo score worsened progressively in the classical PSC group but remained stable in the overlap group during 5 years of follow-up. No deaths were reported in the 7 patients with AIH-PSC overlap versus the 9 deaths (out of 34 patients) in the PSC-only group; this suggests better cumulative survival for AIH-PSC in adults.

In our study, 5 of the 8 patients with the overlapping SC observed on MRC demonstrated an initial good response to corticosteroids and have since remained in clinical remission of their AIH with azathioprine. Another suffered a recent flare of hepatitis after a prolonged remission (medication noncompliance was suspected), whereas the 2 youngest patients (ages 19 and 22), both females who were diagnosed with AIH as teenagers, have not had a sustained remission and are currently being evaluated for a liver transplant. Six of the patients with AIH/SC overlap on MRC have undergone full colonoscopy, and only 1 patient has shown endoscopic and histologic evidence of ulcerative colitis. Of our adult patients with AIH alone (n = 71), 49 (69%) have achieved clinical remission with corticosteroids and are currently maintained on azathioprine and/or low-dose prednisone; 19 patients (27%) are currently asymptomatic on no therapy; and 2 patients (3%) have since died, one from cholangiocarcinoma (not seen on MRC) and another from decompensated liver disease. Only 1 patient has undergone liver transplantation, in contrast to the 2 patients from the group of 8 with SC on MRC. The female preponderance and the younger age at diagnosis of AIH of our patients who were subsequently shown to have overlapping SC on MRC raise the question as to whether this overlap syndrome represents a different clinical entity or a distinct phenotype with respect to PSC alone, which is more common in males. Gregorio et al.5 suggested autoimmune SC to describe this AIH/SC overlap in children and proposed that it is likely the consequence of different pathologies.

As the outcome and management of AIH with overlapping SC seem to differ from those of AIH alone, particularly in children, we postulate that it would be prudent to perform routine surveillance imaging of the biliary tract in all adults with AIH. Knowledge of underlying overlapping SC should lead to appropriate patient counseling, therapy, and follow-up.