Remission, treatment failure, and incomplete response constituted the treatment outcomes, and patients were classified in accordance with these previously published criteria.34 An incomplete response was declared only after at least 6 months of conventional corticosteroid therapy. Recrudescence of disease after remission and withdrawal of medication connoted relapse, and the development of fibrosis with a complete regenerative nodule in sequential liver tissue samples indicated progression to cirrhosis.34 Treatment failure and incomplete response were designated as a poor result. Death from hepatic failure and/or liver transplantation was not included as a poor result unless it occurred during immediate therapy and was categorized as a treatment failure.

Statistical Analyses. Fisher's Exact Test was used to compare dichotomous variables, and the unpaired t test was used to compare differences in the means of continuous variables. Nonparametric variables in independent samples were compared by use of the Mann-Whitney test. A P value of .05 was required for statistical significance. Data are presented as means ± SEM in tables and text.

RESULTS

Frequency of the Variant Syndromes. Of 162 patients with the clinical diagnosis of type 1 autoimmune hepatitis, 11 (7%) satisfied criteria for autoimmune cholangitis and 8 (5%) were redesignated as having autoimmune hepatitis and PBC (table 2). Seven of the 37 patients with PBC (19%) satisfied criteria for autoimmune hepatitis and PBC, and 14 of the 26 patients with PSC (54%) were redesignated as having autoimmune hepatitis and PSC (table 2). Variant syndromes were found more commonly in patients with the original clinical diagnosis of PSC than in patients with the original clinical diagnosis of type 1 autoimmune hepatitis or PBC (table 2). The frequency of the variant syndromes in the entire study population of 225 patients was 18%, and each variant form occurred with similar frequency within this group (table 2).

View This table
table 2. Frequencies of the Variant Syndromes in Autoimmune Chronic Liver Disease

Importantly, patients initially categorized as having type 1 autoimmune hepatitis who were subsequently reclassified as having autoimmune hepatitis and PBC were indistinguishable from those originally categorized as having PBC and then reclassified as having autoimmune hepatitis and PBC. Both groups were similar by age (46 ± 5 years vs. 58 ± 5 years, P = .1), female composition (75% vs. 100%, P= .5), serum levels of aspartate aminotransferase (698 ± 269 U/L vs. 413 ± 150 U/L, P = .4), alkaline phosphatase (461 ± 112 U/L vs. 400 ± 77 U/L, P = .7), bilirubin (6.9 ± 2.2 mg/dL vs. 5.8 ± 3.4 mg/dL, P = .8), and immunoglobulin M (382 ± 69 mg/dL vs. 692 ± 204 mg/dL, P = .1), and each group had comparable mean aggregate scores for autoimmune hepatitis (13 ± 1 vs. 16 ± 1, P = .09).

Distinctive Clinical Features. Patients with autoimmune hepatitis and PBC were distinguished from those with definite autoimmune hepatitis by having lower serum levels of immunoglobulin G, higher serum concentrations of immunoglobulin M, and a lower occurrence of SMA (table 3). Importantly, these patients were indistinguishable from those with definite autoimmune hepatitis by serum aspartate aminotransferase, bilirubin, alkaline phosphatase, and -globulin concentrations (table 3).

View This table
table 3. Features of Variant Syndromes Compared With Definite Autoimmune Hepatitis

Patients with autoimmune hepatitis and PSC had concurrent immune diseases more frequently than patients with definite autoimmune hepatitis (79% vs. 38%, P = .007), but this difference reflected mainly the close association of PSC with inflammatory bowel disease. Serum concentrations of aspartate aminotransferase, -globulin, and immunoglobulin G were lower in these patients than in those with definite autoimmune hepatitis, whereas serum levels of alkaline phosphatase were higher (table 3). Furthermore, autoantibodies occurred less frequently, and HLA-DR4 was detected less often (table 3).

Patients with autoimmune cholangitis were distinguished from those with definite autoimmune hepatitis by lower serum levels of aspartate aminotransferase, -globulin, and immunoglobulin G at accession, higher serum concentrations of alkaline phosphatase, and a lower frequency of SMA (table 3). Importantly, these patients had the lowest mean aggregate score compared with each of the other syndromes, and they were more widely separated from definite autoimmune hepatitis (table 3). Only 1 patient with autoimmune cholangitis had inflammatory bowel disease (Crohn's disease), and findings on cholangiography in this patient were normal. Five other patients underwent retrograde endoscopic cholangiography as part of their routine assessment, and results of these studies were also normal. Eight of the 11 patients had histological changes of ductopenia and/or cholangitis. Three patients had nondiagnostic liver tissue specimens but cholestatic clinical and/or laboratory features.

Treatment Results. Conventional corticosteroid regimens were administered to 12 patients with autoimmune hepatitis and PBC, 9 patients with autoimmune hepatitis and PSC, and 6 patients with autoimmune cholangitis (table 4). Only patients with the overlap syndrome of autoimmune hepatitis and PBC entered remission as commonly as the 86 treated patients with definite autoimmune hepatitis during comparable periods of follow-up. These patients also had a lower frequency of progression to cirrhosis (table 4).

View This table table 4. Results of Corticosteroid Therapy

Remission was less common in patients with autoimmune hepatitis and PSC than in patients with definite autoimmune hepatitis and patients with autoimmune hepatitis and PBC (table 4). Furthermore, the frequency of a poor result was higher in these patients than in each variant syndrome and definite autoimmune hepatitis. Patients with autoimmune hepatitis and PSC also died of liver failure or required liver transplantation more often than patients with classical autoimmune hepatitis (table 4).

Remission was not achieved in any of the patients with autoimmune cholangitis, but the mean duration of treatment and follow-up was short (table 4).

Factors Affecting Results of Corticosteroid Therapy. The frequencies of HLA-DR3 (36% vs. 56%, P = .4) and DR4 (45% vs. 38%, P = .7) were similar in patients with and without remission regardless of variant type. The aggregate score also did not predict treatment response, because patients with remission, treatment failure, and an overall poor result had similar scores (table 5). Furthermore, the number of patients with scores of 15 or greater was similar in each response category (table 5).

View This table
table 5. Features Associated With Corticosteroid Responses in the Variant Syndromes

Remission during corticosteroid therapy was associated with lower serum levels of alkaline phosphatase and higher serum concentrations of -globulin and immunoglobulin G (table 5). Indeed, none of the patients with variant syndromes and serum alkaline phosphatase levels of more than twofold the reference value responded to corticosteroid treatment. The degree of hypergammaglobulinemia did not have a similar discriminatory effect (table 5).

DISCUSSION

Variant forms of immunity-mediated chronic liver disease are common, and they are frequently assimilated into conventional diagnostic categories. In this study, 18% of patients originally designated as having type 1 autoimmune hepatitis, PBC, and PSC had features of autoimmune cholangitis or autoimmune hepatitis and PBC or PSC (table 2). Detection depends on the diagnostic criteria that are applied, and these syndromes lack an established identity, official designation, and assessment algorithm.2,4 Consequently, their prevalence and behavior can vary between studies, and regional results may not be generalizable. A scoring system (table 1) was used in this report mainly to ensure uniformity of assessment and diagnosis. The specificity of this scoring system for autoimmune hepatitis has been established,3 and it provides a template that can be refined with experience or replaced as variant forms are better characterized and diagnostic criteria are codified.

Patients with PSC more commonly had concurrent features of autoimmune hepatitis than patients with PBC (table 2). HLA-DR3, ANA, and/or SMA seropositivity, and extrahepatic immune diseases, including inflammatory bowel disease, are common in both PSC and autoimmune hepatitis.3,14,25 These similarities may have accounted for the high occurrence of features associated with autoimmune hepatitis in the PSC group. The scoring system may overvalue nonspecific clinical features that are common in different diseases and undervalue or neglect to weigh truly discriminative findings, such as cholangiographic changes.3,14 Future studies must refine the diagnostic indices for the PSC variant by demonstrating that this syndrome is different from either primary disease and by tailoring the scoring system to recognize only this form.

Patients with autoimmune hepatitis and PBC or PSC differed from patients with definite autoimmune hepatitis, and the principal areas of difference reflected features particular to the PBC or PSC components of the syndrome (table 3). Autoimmune cholangitis had the lowest net aggregate score, and it was most distantly removed from the diagnosis of autoimmune hepatitis. None of the syndromes, however, could be distinguished from the others by individual findings or aggregate scores (table 3), and their existence as distinct clinical entities could not be established. Larger numbers of patients are required to fully assess the issue of independent existence.

Only patients with autoimmune hepatitis and PBC responded as commonly to corticosteroid therapy as patients with definite autoimmune hepatitis, and these individuals progressed to cirrhosis less frequently than patients with autoimmune hepatitis and PSC and patients with definite autoimmune hepatitis during comparable periods of observation (table 4). Previous studies have indicated that patients with autoimmune hepatitis, AMA seropositivity, and/or histological features of cholangitis frequently respond to corticosteroid therapy,17,18 and our current analysis reconfirms this potential benefit.

Importantly, patients with autoimmune hepatitis and PBC had serum aspartate aminotransferase, bilirubin, alkaline phosphatase, and -globulin levels that were similar to those in patients with definite autoimmune hepatitis, and the frequencies of HLA-DR3 and DR4 were also comparable (table 3). Mean aggregate scores were higher in these patients than those of the other conditions, and although these differences were not statistically significant, they indicated a stronger kinship with autoimmune hepatitis than was evident in the other variants. This greater resemblance may have contributed to the better outcome.

In contrast, none of the patients with autoimmune cholangitis and only 2 of the 10 treated patients with autoimmune hepatitis and PSC entered remission during corticosteroid therapy (table 4). Remission occurred less frequently in these variant conditions than in patients with definite autoimmune hepatitis, and a poor result was especially common in patients with autoimmune hepatitis and PSC (table 4).). Other studies have indicated a similar recalcitrance to corticosteroid therapy in autoimmune cholangitis6,8 and PSC,15,16,25 especially in regard to histological improvement, and this study extends these observations to the variant categories.

Unfortunately, the appropriate therapy for each of the variant syndromes remains uncertain, and treatment strategies are empirical. In this study, responsiveness to corticosteroid therapy was associated with a serum alkaline phosphatase level of less than twofold the reference value (table 5). Improvement during corticosteroid therapy is uncommon in chronic cholestatic liver disease,17,25 and the magnitude of serum alkaline phosphatase elevation may be the most useful laboratory test to differentiate variant syndromes that have predominant cholestatic or hepatocellular features. Further assessment of this laboratory parameter as an index for treatment selection is warranted.

In summary, variant syndromes of immunity-mediated chronic liver disease are common, and they should be sought in all such patients. Only individuals with autoimmune hepatitis and PBC respond well to corticosteroids. Individuals with autoimmune hepatitis and PSC have an especially poor treatment outcome. Normal or near-normal serum alkaline phosphatase levels characterize the treatment responders. The frequency of the variant syndromes and their variable response to corticosteroids justify continued efforts to define their character, establish their validity, codify diagnostic criteria, and determine effective therapies. A greater understanding of shared pathogenic mechanisms and/or host predispositions and a better explanation of unsuccessful treatment in patients with presumed autoimmune hepatitis should result from this effort. Indeed, the recognition and separate categorization of these syndromes may be essential in ensuring accurate depictions of the natural history and prognosis of autoimmune hepatitis.

References