AUTOIMMUNE, CHOLESTATIC AND BILIARY DISEASE Performance Parameters of the Diagnostic Scoring Systems for Autoimmune Hepatitis Albert J. Czaja The diagnostic criteria for autoimmune hepatitis (AIH) have been codified by an international panel, and a revision of the original scoring system based on 12 clinical components has been promulgated. A simplified scoring system has been proposed recently that is based on four clinical components. The goals of this study were to compare the performance parameters of the revised original and the simplified scoring systems and to determine the prowess of each as a diagnostic instrument. Diagnostic scores were determined using each scoring system in 435 patients with diverse chronic liver diseases, including 153 individuals with AIH by codified clinical criteria. The sensitivity, specificity, and predictability of each scoring system for the pretreatment diagnosis of AIH were determined. The revised original scoring system had greater sensitivity for the diagnosis than the simplified scoring system (100% versus 95%), and seven patients diagnosed as AIH using the revised original system were nondiagnostic by the simplified system (5%). The revised original scoring system also ascribed a diagnosis of AIH to 20 of 21 patients with cryptogenic chronic hepatitis, whereas only five patients were similarly classified by the simplified system (95% versus 24%). The simplified system had greater specificity (90% versus 73%) and predictability (92% versus 82%) for AIH than the revised original system, and it more commonly excluded the diagnosis in other diseases with concurrent immune features (83% versus 64%). Conclusion: The revised original scoring system performs better in patients with few or atypical features of AIH, and the simplified system is better at excluding the diagnosis in diseases with concurrent immune manifestations. Each system has attributes that can be exploited. (HEPATOLOGY 2008;48:1540-1548.) T he diagnostic criteria for autoimmune hepatitis (AIH) were codified by an international panel in 1993,1 and these recommendations were updated in 1999.2 They provided clinical guidelines for the definite and probable diagnosis of AIH and a scoring system that allowed assessment of the strength of the diagnosis. The revised original scoring system could be applied before or after corticosteroid treatment, and it was intended as a research tool to allow characterization of study populations within and between clinical studies.2 Abbreviations: AIH, autoimmune hepatitis; ELISA, enzyme-linked immunosorbent assay; HLA, human leukocyte antigen; IgG, immunoglobulin G; ROC, receiver operating characteristic. From the Division of Gastroenterology and Hepatology, Mayo Clinic College of Medicine, Rochester, MN. Received May 9, 2008; accepted June 26, 2008. Address reprint requests to: Albert J. Czaja, M.D., Mayo Clinic College of Medicine, 200 First Street S.W., Rochester, MN 55905. E-mail: [email protected]; fax: 507-284-0538. Copyright © 2008 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/hep.22513 Potential conflict of interest: Nothing to report. 1540 The virtues of the revised original scoring system were that it facilitated objective comparisons between patient populations and accommodated individuals with atypical manifestations. By grading phenotypic, biochemical, serological, epidemiological, genetic, and histological features, the scoring system had a high sensitivity (97% to 100%) for AIH,2 and it could render the diagnosis in individuals who lacked certain classical features (hypergammaglobulinemia or autoantibodies) or who manifested atypical components (antimitochondrial antibodies, cholestasis, or atypical histological features).3,4 The scoring downgrades associated with absent or inconsistent features could be overridden by the strength of other manifestations. The major limitation of the scoring system was its complexity (13 components and 29 possible grades), and its awkwardness as an easily applicable clinical tool.2,3 To resolve these difficulties, the International Autoimmune Hepatitis Group developed a simplified scoring system based on four components and twelve possible grades.5 Unlike the original scoring system, the simplified system was tested in patients from Germany, Japan, Spain, HEPATOLOGY, Vol. 48, No. 5, 2008 Greece, Norway, Brazil, Austria, the United Kingdom, and the United States, and it was also validated in diverse control populations that included cholestatic, viral, metabolic, toxic, and genetic hepatic disorders.5 In this fashion, the universality of the simplified scoring system was established, and its high sensitivity (88%) and specificity (97%) for AIH were demonstrated. The simplified scoring system has not been compared with the revised original scoring system, and its ability to replace this more comprehensive assessment has not been established. Each system was developed with different intentions (research use versus clinical use), and their applications may have different consequences. The comprehensive nature of the revised original scoring system may identify individuals with AIH that would not be detected with the simplified system. Alternatively, the simplified system may include in the diagnosis individuals with atypical features that would have otherwise been excluded by the revised original scoring method. Each system may have different and even complementary roles, and the simplified system may be an expansion rather than a revision of the current diagnostic armamentarium. In this retrospective study, we compare the performance parameters of the revised original2 and simplified5 scoring systems of the International Autoimmune Hepatitis Group in patients satisfying conventional pretreatment clinical criteria for the diagnoses of autoimmune, viral, cholestatic, alcoholic, and metabolic chronic liver diseases, and we determine the concordance and discrepancies between these systems in the same patient populations. In this fashion, we determine the diagnostic prowess of these systems and the complementarity or redundancy of the information they provide. Patients and Methods Study Population Four hundred thirty-five patients who had been evaluated in a uniform fashion by the author were selected for study because they each satisfied conventional clinical criteria for their particular diagnosis at presentation, and they had each been assessed for the features common to both scoring systems, including serum immunoglobulin G (IgG) levels, conventional autoantibodies by indirect immunofluorescence, serological markers for hepatitis B and hepatitis C viruses, and histological examination.2,5 One hundred fifty-three patients (35%) were diagnosed as having definite or probable AIH according to the codified clinical criteria promulgated by the International Autoimmune Hepatitis Group in 19992 and reflected in the practice guidelines of the American Association for the Study of Liver Diseases in 2002.6 These patients were assessed at presentation, and their diagnosis was based on CZAJA 1541 the codified clinical (not scoring) criteria espoused in 1999.2 One hundred thirty-nine patients (91%) received conventional corticosteroid regimens according to protocols published previously,6 including prednisone alone (51 patients) or a lower dose of prednisone in conjunction with azathioprine (88 patients). Four patients had received investigational therapies, and 10 patients had not been treated at the time of diagnosis. The treatment outcomes of remission or treatment failure were assessed by applying the definitions reported.6 One hundred twenty patients (28%) were diagnosed with conventional clinical criteria as having chronic hepatitis C7; 18 patients (4%) were diagnosed as having chronic hepatitis B7,8; 46 patients (10%) were diagnosed as having nonalcoholic fatty liver disease9,10; 42 patients (10%) were diagnosed as having primary biliary cirrhosis11; 24 patients (6%) were diagnosed as having primary sclerosing cholangitis12; 21 patients (5%) were diagnosed as having cryptogenic chronic hepatitis13-16; and 11 patients (2%) were diagnosed as having chronic alcoholic liver disease.17 These 435 patients were further classified into the categories of autoimmune (n ⫽ 153), viral (n ⫽ 138), metabolic (n ⫽ 46), cholestatic (n ⫽ 66), cryptogenic (n ⫽ 21), and alcoholic (n ⫽ 11) liver disease. The study was approved by the Institutional Review Board of the Mayo Clinic. Clinical and Laboratory Assessments Clinical examinations and conventional laboratory tests of liver inflammation and function had been performed at presentation and before the institution of specific therapy in all patients in accordance with a published protocol.6 Concurrent extrahepatic disorders of an immune nature had been sought in a systematic fashion as described.6 Serum protein electrophoresis had been assessed via immunonephelometry in 431 patients (99%), and serum IgG concentrations had been determined via a similar method in all patients. Smooth muscle antibodies18 and antinuclear antibod19 ies had been assessed via indirect immunofluorescence on murine tissue sections in all patients. Antibodies to liver kidney microsome type 1 had been evaluated via indirect immunofluorescence on murine tissue sections in 402 patients (92%),20 and antimitochondrial antibodies had been assessed in all but one patient (99%) via indirect immunofluorescence (n ⫽ 428)21 or via enzyme-linked immunosorbent assay (ELISA) (n ⫽ 6).22 All patients had undergone at least three of the conventional serological tests for autoantibodies, and 403 patients (93%) had been evaluated using all four tests. Hepatitis B surface antigen (Ortho Diagnostic Systems, Inc., Raritan, NJ) and antibodies to hepatitis C 1542 CZAJA HEPATOLOGY, November 2008 Table 1. Revised Original Pretreatment Scoring System of the International Autoimmune Hepatitis Group* Component Sex AP:AST (or ALT) ratio ␥-globulin or IgG level above normal ANA, SMA, or anti-LKM1 titers AMA Viral markers Drugs Alcohol Result Points Component Result Points Female ⬎3 ⬍1.5 ⬎2.0 1.5–2.0 1.0–1.5 ⬍1.0 ⬎1:80 1:80 1:40 ⬍1:40 ⫹2 ⫺2 ⫹2 ⫹3 ⫹2 ⫹1 0 ⫹3 ⫹2 ⫹1 0 HLA Immune disease DR3 or DR4 Thyroiditis, colitis, others ⫹1 ⫹2 Other markers Anti-SLA, actin, LC1, pANCA ⫹2 Histological features Positive Positive Negative Yes No ⬍25g/day ⬎60g/day ⫺4 ⫺3 ⫹3 ⫺4 ⫹1 ⫹2 ⫺2 Treatment response Interface hepatitis Plasmacytic Rosettes None of above Biliary changes Other features Complete Relapse ⫹3 ⫹1 ⫹1 ⫺5 ⫺3 ⫺3 ⫹2 ⫹3 Pretreatment aggregate score: Definite diagnosis >15 Probable diagnosis 10 –15 Abbreviations: AP, alkaline phosphatase; ALT, alanine aminotransferase; AMA, antimitochondrial antibody; ANA, antinuclear antibody; anti-LC1, antibody to liver cytosol type 1; anti-LKM1, antibodies to liver/kidney microsome type 1; anti-SLA, antibody to soluble liver antigen; AST, aspartate aminotransferase; HLA, human leukocyte antigen; IgG, immunoglobulin G; pANCA, perinuclear anti-neutrophil cytoplasmic antibodies; SMA, smooth muscle antibody. *Adapted from Alvarez et al.2 virus (Abbott Laboratories, North Chicago, IL) had been assessed in all patients via second-generation ELISA. Histological Assessments Liver tissue examinations had been performed at accession in all patients, and the liver specimens have been examined by members of the liver pathology working group at the Mayo Clinic. The pathological diagnoses were rendered in accordance with pre-established criteria.23,24 Our previous validation studies have indicated that the reproducibility of the histological interpretation was 94%.23,25 Human Leukocyte Antigen Determinations Four hundred ten patients (94%) had been evaluated for the class II (DR locus) human leukocyte antigens (HLA) by polymerase chain reaction with sequence-specific oligonucleotide probes (n ⫽ 187),26 restriction fragment length polymorphism (n ⫽ 217),27 or microlymphocytotoxicity (n ⫽ 6).28 Scoring Systems The revised original scoring system of the International Autoimmune Hepatitis Group and the simplified system were applied to all patients.2,5 The 12 components of the revised original scoring system that were applicable to the pretreatment evaluation (Table 1) were graded in the 410 patients who had undergone HLA testing. The 25 pa- tients untested for HLA were graded for the other 11 components that were applicable to the pretreatment evaluation (Table 1). Only the pretreatment criteria for the diagnosis of AIH were applied, and a score of 10 to 15 points supported a probable diagnosis and a score of ⬍15 points supported a definite diagnosis by this revised original method (Table 1).2 The four components of the simplified scoring system were graded in all patients (Table 2). A score of 6 points constituted a probable diagnosis of AIH using the simplified system, and a score of ⱖ7 points constituted a definite diagnosis (Table 2).5 Statistical Analyses Sensitivity, specificity, and predictability were the performance parameters for diagnosis that were assessed for each scoring system. The clinical diagnoses based on codified criteria for AIH pretreatment2 and on conventional clinical criteria for the other chronic liver diseases7-17 were the gold standards of diagnosis. Sensitivity. Sensitivity reflected the number of patients with definite or probable AIH by the codified clinical criteria who were diagnosed similarly by the scoring system under study (true positives) divided by the sum of true positives plus the number of patients with the disease by the codified clinical criteria who were undiagnosed by the scoring system under study (false negatives). Specificity. Specificity reflected the number of pa- HEPATOLOGY, Vol. 48, No. 5, 2008 CZAJA Table 2. Simplified Scoring System of the International Autoimmune Hepatitis Group* Component Autoantibodies ANA or SMA ANA or SMA Antibodies to liver kidney microsome type 1 Antibodies to soluble liver antigen Absent autoantibodies Immunoglobulin level Immunoglobulin G Histological Findings Morphological features of AIH Viral Disease Absence of viral hepatitis Result Points ⱖ1:40 ⱖ1:80 ⱖ1:40 Positive None ⫹1 ⫹2 ⫹2 ⫹2 0 ⬎UNL ⬎1.1 ULN Normal ⫹1 ⫹2 0 Compatible Typical Incompatible ⫹1 ⫹2 0 No viral markers Viral markers prsent ⫹2 0 ⱖ7 6 1543 nal scoring system and the simplified scoring system (Analyze-it Software, Ltd., Leeds, United Kingdom). Results Pretreatment Aggregate Score: Definite diagnosis Probable diagnosis Abbreviations: ANA, antinuclear antibodies; SMA, smooth muscle antibodies; ULN, upper limit of normal range. *Adapted from Hennes et al.5 tients without AIH by the codified clinical criteria who were also discounted by the scoring system under study (true negatives) divided by the sum of true negatives plus the number of patients without AIH by the codified clinical criteria who were diagnosed with the disease by the scoring system under study (false positives). Predictability. Predictability reflected the number of true positives and true negatives divided by the entire study population. Positive predictability reflected the number of true positives divided by the sum of true positives and false positives. Negative predictability reflected the number of true negatives divided by the sum of true negatives and false negatives. Receiver Operating Characteristic Curves. Receiver operating characteristic (ROC) curves were plotted to illustrate the sensitivity and specificity of the revised origi- Comparison of Scoring Systems for Autoimmune Hepatitis. The revised original scoring system of the International Autoimmune Hepatitis Group graded all patients who satisfied the codified clinical criteria for AIH as having either definite (n ⫽ 140) or probable (n ⫽ 13) AIH. In contrast, the simplified scoring system made the diagnosis of definite (n ⫽ 132) and probable (n ⫽ 14) AIH less frequently (95% versus 100%) (Table 3). The frequency of a definite diagnosis was greater by the revised original scoring system than by the simplified scoring system (92% versus 86%), and seven patients (5%) who had been graded as definite (n ⫽ 2) or probable (n ⫽ 5) AIH according to the revised original scoring system had nondiagnostic scores using the simplified scoring system (Table 3). Of the 140 patients graded as definite AIH using the revised original scoring system, 13 patients (9%) were classified as probable (n ⫽ 11) or no AIH (n ⫽ 2) using the simplified system. Of the 13 patients with probable AIH using the revised original scoring system, 10 patients were classified as definite (n ⫽5) or no AIH (n ⫽ 5) using the simplified system. The scoring systems rendered concordant diagnoses of definite (n ⫽ 127) or probable AIH (n ⫽ 3) in 130 patients (85%). Discordant diagnoses of AIH were rendered in 23 patients (15%). Scoring Discrepancies for Autoimmune Hepatitis. Sixteen of the 23 patients with discrepant diagnoses for AIH had either been scored as definite AIH using the revised original scoring system and probable AIH using the simplified scoring system (n ⫽ 11) or definite AIH using the simplified system and probable AIH using the revised original scoring system (n ⫽ 5). These differences related to the points awarded or subtracted for features Table 3. Comparison of Clinical Diagnoses and Scoring Diagnoses Using Each System Diagnosis by Scoring System, n (%) Definite AIH Probable AIH Nondiagnostic Clinical Diagnosis Original* Simplified Original* Simplified Original* Simplified AIH (n ⫽ 153) Cryptogenic hepatitis (n ⫽ 21) Chronic alcoholic liver disease (n ⫽ 11) Chronic hepatitis B (n ⫽ 18) Chronic hepatitis C (n ⫽ 120) Primary biliary cirrhosis (n ⫽ 42) Primary sclerosing cholangitis (n ⫽ 24) Nonalcoholic fatty liver disease (n ⫽ 46) 140 (92) 1 (5) 0 (0) 0 (0) 0 (0) 0 (0) 2 (8) 1 (2) 132 (86) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 2 (8) 0 (0) 13 (8) 19 (90) 4 (36) 6 (33) 32 (27) 2 (5) 7 (29) 3 (7) 14 (9) 5 (24) 0 (0) 2 (11) 10 (8) 6 (14) 3 (13) 1 (2) 0 1 (5) 7 (64) 12 (67) 88 (73) 40 (95) 15 (63) 42 (91) 7 (5) 16 (76) 11 (100) 16 (89) 110 (92) 36 (86) 19 (79) 45 (98) *Original ⫽ revised original scoring system (Table 1). 1544 CZAJA HEPATOLOGY, November 2008 Table 4. Features Associated with Diagnostic Discrepancies Between Scoring Systems for Autoimmune Hepatitis Number of Scoring Adjustments Accounting for Discrepant Diagnoses in 23 Patients Features Unrecognized or Scored Less via Simplified Scoring System Female sex Concurrent immune diseases AP/AST ratio ⬎1.5 and ⬍3 AP/AST ratio ⬍1.5 Abnormal serum ␥-globulin and normal serum IgG levels High-titer autoantibodies AMA present Exposure to possible hepatotoxic medication HLA DRB1*03, DRB1*04, or both Atypical histological features Original Scoring System* Upgrades Original Scoring System* Downgrades 16 6 — 21 — — 2 — 8 14 — — — 3 — 1 13 — — 1 Abbreviations: AP, alkaline phosphatase; AMA, antimitochondrial antibodies; AST, aspartate aminotransferase; IgG, immunoglobulin G. *Original scoring system ⫽ revised original scoring system (Table 1). graded in the revised original scoring system but not in the simplified scoring system (Table 4). Points awarded by the revised original scoring system for female sex (n ⫽ 9), abnormal serum ␥-globulin level despite normal serum IgG level (n ⫽ 6), high-titer autoantibodies (n ⫽ 10), ratio of serum alkaline phosphatase to aspartate aminotransferase levels less than 1.5 (n ⫽ 11), concurrent immune diseases (n ⫽ 3), and presence of HLA DRB1*03, DRB1*04 or both (n ⫽ 5) were the bases for ascribing a definite diagnosis to the 11 patients graded as having a probable diagnosis using the simplified scoring system. Points subtracted or not awarded by the revised original scoring system for male sex (n ⫽ 3), atypical histological features (n ⫽ 1), antimitochondrial antibodies (n ⫽ 3), ratio of serum alkaline phosphatase to aspartate aminotransferase levels greater than 1.5 and less than 3.0 (n ⫽ 2), lack of concurrent immune diseases (n ⫽ 5), exposure to possible hepatotoxic medication (n ⫽ 1), and lack of classical HLA markers (n ⫽ 2) were the bases for ascribing a probable diagnosis to the five patients graded as having a definite diagnosis using the simplified scoring system. Similarly, the discrepant diagnoses in the seven patients graded definite or probable AIH using the revised original system and nondiagnostic for AIH using the simplified system related to features not counted or graded as strongly using the simplified system, including female sex (n ⫽ 5), concurrent immune diseases (n ⫽ 3), abnormal serum ␥-globulin levels despite normal serum IgG levels (n ⫽ 2), high-titer autoantibodies (n ⫽ 1), ratio of serum alkaline phosphatase to aspartate aminotransferase levels less than 1.5 (n ⫽ 7), and the presence of HLA DRB1*0301, DRB1*0401, or both (n ⫽ 5). The nature and frequency of the features that resulted in diagnostic upgrades or downgrades in the revised original scoring system and that contributed to the diagnostic discrepancies with the simplified scoring system are detailed in Table 4. Treatment Outcomes of Probable and Definite Autoimmune Hepatitis. In the 139 treated patients with definite (n ⫽ 131) or probable (n ⫽ 8) AIH using the revised original scoring criteria, remission (68% versus 75%, P ⬍ 0.9) and treatment failure (16% versus 0%, P ⫽ 0.6) occurred with similar frequencies. In the 134 treated patients with definite (n ⫽ 122) or probable (n ⫽ 12) AIH using the simplified scoring criteria, the frequencies of remission (70% versus 50%, P ⫽ 0.2) and treatment failure (15% versus 25%, P ⫽ 0.4) were also similar. Of the five treated patients who had nondiagnostic scores for AIH using the simplified scoring system, four entered remission, and one continues on therapy. Comparison of Scoring Systems for Excluding Other Diagnoses. Neither the revised original nor the simplified scoring system graded patients with chronic alcoholic liver disease, chronic hepatitis B, chronic hepatitis C, or primary biliary cirrhosis as having definite AIH (Table 3). Both systems graded two patients with primary sclerosing cholangitis as definite AIH, and the original scoring system made the same designation in one patient with cryptogenic chronic hepatitis and one patient with nonalcoholic fatty liver disease. Probable diagnoses of AIH were rendered more commonly using the revised original scoring system in all diagnostic categories except for primary biliary cirrhosis. The revised original scoring system graded 20 of 21 patients with cryptogenic chronic hepatitis (95%) as having definite (n ⫽ 1) or probable (n ⫽ 19) AIH, whereas the simplified scoring system designated only five patients in the probable category (24%) (Table 3). Of 282 patients with diagnoses other than AIH using clinical criteria, 166 (59%) had concurrent immune features, including autoantibodies (n ⫽ 50), abnormally increased serum IgG level (n ⫽ 71), or both (n ⫽ 45). The revised original scoring system graded 60 of these patients (36%) as having either definite (n ⫽ 3) or probable (n ⫽ 57) AIH, whereas the simplified system designated only 29 patients (17%) in this fashion. The simplified scoring system more commonly excluded the diagnosis of AIH in patients with other diagnoses and concurrent immune manifestations than the revised original scoring system (83% versus 64%). HEPATOLOGY, Vol. 48, No. 5, 2008 CZAJA 1545 Table 5. Diagnosis of Autoimmune Hepatitis in Other Disease Categories by Original and Simplified Scoring Systems Diagnosis by Scoring System, n (%) Definite AIH Probable AIH Nondiagnostic Clinical Diagnostic Category Original* Simplified Original* Simplified Original* Simplified Non-AIH (n ⫽ 282) Chronic viral hepatitis (n ⫽ 138) Chronic nonviral liver disease (n ⫽ 144) Chronic cholestatic liver disease (n ⫽ 66) Chronic metabolic liver disease (n ⫽ 46) 4 (1) 0 (0) 4 (3) 2 (3) 1 (2) 2 (1) 0 (0) 2 (1) 2 (3) 0 (0) 73 (26) 38 (28) 35 (24) 9 (14) 3 (7) 27 (10) 12 (9) 15 (11) 9 (14) 1 (2) 205 (73) 100 (72) 105 (73) 55 (83) 42 (92) 253 (89) 126 (91) 127 (88) 55 (83) 45 (98) *Original ⫽ revised original scoring system (Table 1). Definite diagnoses of AIH were unusual with each scoring system when clinical categories of chronic liver disease rather than individual clinical diagnoses were assessed (Table 5). In contrast, probable diagnoses were more common when using the revised original scoring system versus the simplified scoring system in patients with different categories of chronic liver disease, especially in the patients with chronic viral hepatitis. The only exception was in the patients with chronic cholestatic liver disease in whom the performance parameters of each scoring system were identical. Performance Parameters for Diagnosing Autoimmune Hepatitis. Both scoring systems had high sensitivity, specificity, and predictability for diagnosing definite or probable AIH when compared with the codified clinical criteria for diagnosis (Table 6). The revised original scoring system had complete sensitivity and negative predictability for AIH, but it was less specific for the diagnosis than the simplified scoring system and it had a lower positive predictability (Table 6). The advantages of the simplified scoring system reflected its better performance in patients with diagnoses other than AIH (Table 3). Scores higher than 10 in the revised original scoring system (Fig. 1A, right to left on the ROC curve) and higher than 6 in the simplified scoring system (Fig. 1B, right to left on the ROC curve) increased the specificity of the scoring system but decreased its sensitivity. Table 6. Performance Parameters of Scoring Systems for Diagnosis of Autoimmune Hepatitis Performance Parameters Original Scoring System,* n (%) Simplified Scoring System, n (%) Sensitivity Specificity Predictability Positive predictability Negative predictability 153/153 (100) 205/282 (73) 358/435 (82) 153/230 (67) 205/205 (100) 146/153 (95) 253/282 (90) 399/435 (92) 146/175 (83) 253/260 (97) *Original scoring system ⫽ revised original scoring system (Table 1). Fig. 1. ROC curves for (A) the revised original scoring system and (B) the simplified scoring system in the 435 patients. The point on the ROC curve that has been proposed as the minimum pretreatment diagnostic score for probable AIH is shown for each system. Left to right movement along the ROC curve decreases the diagnostic score, thereby increasing the sensitivity and decreasing the specificity of the score. 1546 CZAJA Discussion This study indicates that each scoring system proposed by the International Autoimmune Hepatitis Group has high sensitivity and specificity for the diagnosis of definite or probable AIH and that the simplified scoring system has superior specificity and predictability for the disease (Table 6). The scoring systems, however, are not fully interchangeable, and in certain clinical circumstances, one system may be more valuable than the other. Autoimmune hepatitis does not have pathognomonic clinical, laboratory, or histological features, and other chronic liver diseases—including alcoholic, viral, cholestatic, drug-induced, and metabolic disorders—may have similar findings (Table 3).29-33 The higher specificity and predictability of the simplified scoring system for AIH may reduce the diagnostic confusion that can be associated with the trappings of hypergammaglobulinemia and autoantibody positivity.7 Furthermore, scores for the ratio of the serum alkaline phosphatase level to the aspartate aminotransferase level, serum ␥-globulin concentration, serum IgG level, autoantibody titer, and individual histological features may be more reflective of the severity than the nature of the liver disease (Table 1). These features are intrinsic to the revised original scoring system,2 and they may render a greater number of probable diagnoses of AIH in diseases that are etiologically distinct than the simplified system.34 Mild forms of AIH may not be as well recognized via the revised original scoring system, whereas its application to all patients with chronic liver disease may promote designations of an overlap syndrome. In the former circumstance, the clinical diagnosis may not be supported by the scoring method,35 and in the latter circumstance, a diagnosis may be rendered that lacks a codified definition or treatment strategy.29,33,34 Alternatively, the simplified scoring system may underestimate the diagnosis of AIH in patients who have few or atypical features of the disease. Seven of the 153 patients (5%) designated as having AIH using the revised original scoring system were not graded as AIH via the simplified scoring system (Table 3). Furthermore, 20 of 21 patients who had been designated as cryptogenic chronic hepatitis were recognized as AIH using the revised original scoring system in contrast to only 5 patients (24%) graded using the simplified scoring system (Table 3). The revised original scoring system ensures the application of a comprehensive diagnostic template, and in this fashion, it may render the diagnosis of AIH in patients unrecognized via the simplified scoring system or designated as cryptogenic chronic hepatitis using conventional clinical criteria (Table 3). HEPATOLOGY, November 2008 Discrepancies in diagnosis have clinical relevance only if they mask the outcome of the disease or affect therapeutic actions. Not all patients with AIH have immediately life-threatening disease or require corticosteroid treatment, and the underestimation of the diagnosis of AIH in a patient with few or atypical features may have no clinical consequence.36 The seven patients who were not diagnosed with AIH via the simplified scoring system may not have suffered from the misclassification. Similarly, the consequences of misclassifying a patient as definite or probable AIH is unlikely to affect prognosis or treatment since the therapeutic decision is based on the clinical manifestations of disease severity.6 Furthermore, the response to corticosteroid therapy provides reassurance about the accuracy of the original diagnosis, and a treatment trial is encouraged by the revised original scoring system.2 The threshold of tolerance for a misclassification by the scoring systems is unknown, and until the natural history of nonclassical AIH and the prognostic distinctions between definite and probable AIH are fully clarified, an error in diagnosis via any scoring system cannot be rationalized. In this study, the patients with probable or definite diagnoses of AIH via each scoring system responded as well to corticosteroid therapy, and this finding supported the accuracy of the codified clinical criteria of the International Autoimmune Hepatitis Group for identifying patients with the diagnosis of AIH, strengthened the gold standard against which each scoring system was tested in this study, and suggested that distinctions between probable and definite diagnoses of AIH did not impact on treatment outcome. Four of the five treated patients with nondiagnostic scores for AIH via the simplified scoring system responded well to corticosteroid treatment, and this finding supported the value of applying the revised original scoring system to patients with few or atypical findings at presentation. Another discrepancy between the scoring systems was in the reclassification of cryptogenic chronic hepatitis as AIH (Table 3). Many patients with seronegative chronic hepatitis respond as well to corticosteroid therapy as patients with classical AIH,15,16 and any denial or delay in the institution of effective treatment in such patients may affect their outcome. The ability of the revised original scoring system to support the diagnosis of AIH in patients who lack the conventional serological markers of the disease may encourage treatment and impact on outcome. No scoring system can supplant clinical judgment in making a diagnosis or in instituting proper therapy, but a scoring system that sharpens clinical judgment— especially in difficult cases—must benefit the patient, and the revised original scoring system may have this advantage. HEPATOLOGY, Vol. 48, No. 5, 2008 Each scoring system was developed for different reasons, and our study indicates that they each contribute in different ways to the diagnosis of AIH. The revised original scoring system was developed to ensure comparable study populations in clinical trials, and it renders an aggregate score that reflects all aspects of the clinical syndrome. Its capacity in this regard remains unchallenged, and it agrees most closely with the codified clinical criteria for the diagnosis of AIH (Table 3). Our findings indicate that the revised original scoring system is most useful in the evaluation of patients with few or no features of AIH but in whom other diagnostic considerations have been excluded. In this fashion, it can support the clinical diagnosis in patients who lack classical features or who have atypical manifestations. The simplified scoring system was developed to provide a convenient tool that can be easily applied to support the clinical diagnosis, and it performs well against the revised original scoring system in this regard (Table 6). Its major values are its diagnostic specificity, especially in excluding patients with etiologically distinctive disease who have concurrent immune trappings, and the ease of its application in the clinical setting. Until the diagnostic relevance of overlap or variant syndromes of AIH is codified, the simplified system will minimize diagnostic confusion and focus the treatment decision. Its simplicity and accuracy make it a feasible adjunct to everyday clinical practice. Both scoring systems exist to support, not supersede, the clinical diagnosis, and their existence can be justified only by this capacity. Their clinical relevance relates to their ability to strengthen diagnostic confidence— especially in individuals with confusing clinical phenotypes— and this ability relates to the scoring attributes particular to each system. Future studies must compare the performance parameters in patients with different levels of disease severity and with autoantibodies detected via ELISA instead of indirect immunofluorescence. As serological assays based on ELISA replace those based on indirect immunofluorescence in many centers, each scoring system must continue to be updated and revised. In conclusion, both scoring systems of the International Autoimmune Hepatitis Group have high sensitivity and specificity for the diagnosis of AIH, and the simplified scoring system has superior specificity and predictability. The systems are not interchangeable, and each may have value in certain clinical situations. The revised original scoring system has greater value in diagnosing patients with few or atypical features of AIH, especially in patients with cryptogenic or autoantibody-negative chronic hepatitis, and the simplified scoring system can exclude the diagnosis more frequently in patients with CZAJA 1547 etiologically distinctive disease who have concurrent immune manifestations. Each system has been validated against codified clinical criteria for the diagnosis, and each can support but not supersede the clinical diagnosis. References 1. Johnson PJ, McFarlane IG, Alvarez F, Bianchi FB, Bianchi L, Burroughs A, et al. Meeting report. International Autoimmune Hepatitis Group. HEPATOLOGY 1993;18:998-1005. 2. Alvarez F, Berg PA, Bianchi FB, Bianchi L, Burroughs AK, Cancado EL, et al. International Autoimmune Hepatitis Group report: review of criteria for diagnosis of autoimmune hepatitis. J Hepatol 1999;31:929-938. 3. Czaja AJ, Carpenter HA. Validation of a scoring system for the diagnosis of autoimmune hepatitis. Dig Dis Sci 1996;41:305-314. 4. Boberg KM, Fausa O, Haaland T, Holter E, Mellbye OJ, Spurkland A, et al. Features of autoimmune hepatitis in primary sclerosing cholangitis: an evaluation of 114 primary sclerosing cholangitis patients according to a scoring system for the diagnosis of autoimmune hepatitis. HEPATOLOGY 1996;23:1369-1376. 5. Hennes EM, Zeniya M, Czaja AJ, Pares A, Dalekos GN, Krawitt EL, et al. Simplified diagnostic criteria for autoimmune hepatitis. HEPATOLOGY 2008;48:169-176. 6. Czaja AJ, Freese DK. Diagnosis and treatment of autoimmune hepatitis. HEPATOLOGY 2002;36:479-497. 7. Desmet VJ, Gerber M, Hoofnagle JH, Manns M, Scheuer PJ. Classification of chronic hepatitis: diagnosis, grading and staging. HEPATOLOGY 1994;19:1513-1520. 8. Lok AS, McMahon BJ. Chronic hepatitis B. HEPATOLOGY 2007;45:507539. 9. Clark JM, Brancati FL, Diehl AM. Nonalcoholic fatty liver disease. Gastroenterology 2002;122:1649-1657. 10. Neuschwander-Tetri BA, Caldwell SH. Nonalcoholic steatohepatitis: summary of an AASLD Single-Topic Conference. HEPATOLOGY 2003;37: 1202-1219. 11. Taal BG, Schalm SW, ten Kate FW, Hermans J, Geertzen RG, Feltkamp BE. Clinical diagnosis of primary biliary cirrhosis: a classification based on major and minor criteria. Hepatogastroenterology 1983;30:178-182. 12. Wiesner RH, LaRusso NF. Clinicopathologic features of the syndrome of primary sclerosing cholangitis. Gastroenterology 1980;79:200-206. 13. Czaja AJ. Chronic active hepatitis: the challenge for a new nomenclature. Ann Intern Med 1993;119:510-517. 14. Czaja AJ. The variant forms of autoimmune hepatitis. Ann Intern Med 1996;125:588-598. 15. Czaja AJ, Carpenter HA, Santrach PJ, Moore SB, Homburger HA. The nature and prognosis of severe cryptogenic chronic active hepatitis. Gastroenterology 1993;104:1755-1761. 16. Gassert DJ, Garcia H, Tanaka K, Reinus JF. Corticosteroid-responsive cryptogenic chronic hepatitis: evidence of seronegative autoimmune hepatitis. Dig Dis Sci 2007;52:2433-2437. 17. Dunn W, Angulo P, Sanderson S, Jamil LH, Stadheim L, Rosen C, et al. Utility of a new model to diagnose an alcohol basis for steatohepatitis. Gastroenterology 2006;131:1057-1063. 18. Czaja AJ, Cassani F, Cataleta M, Valentini P, Bianchi FB. Frequency and significance of antibodies to actin in type 1 autoimmune hepatitis. HEPATOLOGY 1996;24:1068-1073. 19. Czaja AJ, Cassani F, Cataleta M, Valentini P, Bianchi FB. Antinuclear antibodies and patterns of nuclear immunofluorescence in type 1 autoimmune hepatitis. Dig Dis Sci 1997;42:1688-1696. 20. Czaja AJ, Manns MP, Homburger HA. Frequency and significance of antibodies to liver/kidney microsome type 1 in adults with chronic active hepatitis. Gastroenterology 1992;103:1290-1295. 21. Czaja AJ, Carpenter HA, Manns MP. Antibodies to soluble liver antigen, P450IID6, and mitochondrial complexes in chronic hepatitis. Gastroenterology 1993;105:1522-1528. 1548 CZAJA 22. Montano-Loza AJ, Carpenter HA, Czaja AJ. Frequency, behavior and prognostic implications of antimitochondrial antibodies in type 1 autoimmune hepatitis. J Clin Gastroenterol. In press. 23. Czaja AJ, Carpenter HA. Sensitivity, specificity and predictability of biopsy interpretations in chronic hepatitis. Gastroenterology 1993;105:1824-1832. 24. Czaja AJ, Carpenter HA. Optimizing diagnosis from the medical liver biopsy. Clin Gastroenterol Hepatol 2007;5:898-907. 25. Soloway RD, Baggenstoss AH, Schoenfield LJ, Summerskill WHJ. Observer error and sampling variability tested in evaluation of hepatitis and cirrhosis by liver biopsy. Am J Dig Dis 1975;20:1087-1090. 26. Czaja AJ, Strettell MDJ, Thomson LJ, Santrach PJ, Moore SB, Donaldson PT, et al. Associations between alleles of the major histocompatibility complex and type 1 autoimmune hepatitis. HEPATOLOGY 1997;25:317-323. 27. Mytilineos J, Scherer S, Opelz G. Comparison of RFLP-DR beta and serological HLA-DR typing in 1500 individuals. Transplantation 1990; 50:870-873. 28. Terasaki P, McClelland JD, Park MS, McCurdy B. Microdroplet lymphocyte cytotoxicity test. In: DHEW Publication (NIH) 74-545. Manual of tissue typing techniques. Washington, DC: US Government Printing Office, 1973:54-61. 29. Czaja AJ. Frequency and nature of the variant syndromes of autoimmune liver disease. HEPATOLOGY 1998;28:360-365. HEPATOLOGY, November 2008 30. Czaja AJ, Carpenter HA, Santrach PJ, Moore SB. Genetic predispositions for immunological features in chronic liver diseases other than autoimmune hepatitis. J Hepatol 1996;24:52-59. 31. Loria P, Lonardo A, Leonardi F, Fontana C, Carulli L, Verrone AM, et al. Non-organ-specific autoantibodies in nonalcoholic fatty liver disease: prevalence and correlates. Dig Dis Sci 2003;48:2173-2181. 32. Adams LA, Lindor KD, Angulo P. The prevalence of autoantibodies and autoimmune hepatitis in patients with nonalcoholic fatty liver disease. Am J Gastroenterol 2004;99:1316-1320. 33. Czaja AJ. Overlap syndrome of primary biliary cirrhosis and autoimmune hepatitis: a foray across diagnostic boundaries. J Hepatol 2006;44:251252. 34. Talwalkar JA, Keach JC, Angulo P, Lindor KD. Overlap of autoimmune hepatitis and primary biliary cirrhosis: an evaluation of a modified scoring system. Am J Gastroenterol 2002;97:1191-1197. 35. Czaja AJ, Bianchi FB, Carpenter HA, Krawitt EL, Lohse AW, Manns MP, et al. Treatment challenges and investigational opportunities in autoimmune hepatitis. HEPATOLOGY 2005;41:207-215. 36. Feld JJ, Dinh H, Arenovich T, Marcus VA, Wanless IR, Heathcote EJ. Autoimmune hepatitis: effect of symptoms and cirrhosis on natural history and outcome. HEPATOLOGY 2005;42:53-62.
