Ethics
This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Kagawa University, Faculty of Medicine (Heisei-30-151) [11]. Informed consent was obtained from patients for analysis of clinical data. For patients who had died and had no relatives listed in their clinical records, we provided opt-out methods for the relatives of the dead participants by publishing a summary of this study on our university website [12, 13].
Study design
Patients were retrospectively assigned to definite AIH or probable AIH based on IAIHG scoring system. Survival rates of two groups were compared by Kaplan-Meier curves. Cox proportional hazard model was performed to identify independent risk factors for survival.
Patients
Among 1957 patients who underwent percutaneous liver biopsy examinations between November 1, 1987 and December 31, 2018, those who were suspected of having AIH, primary biliary cholangitis (PBC), or nonalcoholic steatohepatitis (NASH) were enrolled in this retrospective cohort study. Clinical data of the patients were reviewed and scored by the IAIHG criteria as described below. Definite and probable AIH patients were analyzed further.
Clinical data
The following clinical data were extracted from the patients’ medical records: age, sex, history of alcohol consumption and internal medicines, patient complications, and family history of autoimmune diseases. For laboratory data, platelet count, aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), total bilirubin (T-Bil), serum albumin (Alb), and immunoglobulin G (IgG) levels were recorded. T-Bil (mg/dl) was converted to T-Bil (µmol/l) according to the following equation: T-Bil (mg/dl) × 17.2. Serological evaluation was based on anti-nuclear antibody (ANA), anti-smooth muscle antibody (SMA), anti-type-1 liver-kidney microsomal (LKM-1) antibody, anti-mitochondrial antibody (AMA), and anti-mitochondrial M2 antibody [14]. Other autoantibodies and human leukocyte antigen (HLA) typing were also referred to when the data were relevant. Albumin-bilirubin (ALBI) score was calculated based on a calculation from a previous report: (Log 10 T-Bil (µmol/l) × 0.66 + Alb (g/l) × (-0.085) [15]. Fibrosis-4 index (Fib-4 index), a conventional liver fibrosis index, was calculated using the following equation: age × AST (U/l) / (Plt (109/l) × √ALT (U/l)) [16].
Histopathological analysis
Pathological findings of liver biopsy specimens were evaluated according to IAIHG criteria [7]. In the current analysis, hepatic steatosis equal to or more than 5% accompanied by hepatocyte ballooning was designated as the prominent feature of NASH [17]. Pathological evaluation was performed by an experienced pathologist who specialized in liver pathology.
Diagnosis of autoimmune hepatitis
According to IAIHG criteria revised in 1999, diagnosis of pre-treatment definite AIH was determined based on a score above 15 [7]. Pre-treatment probable AIH was diagnosed in patients with scores ranging between 10 and 15. Post-treatment definite or probable AIH was diagnosed as described in the IAIHG criteria.7 To calculate diagnostic scores, 35 U/l was applied as the upper limit of the normal (ULN) AST values, and 40 U/l for ALT and 17.0 g/l for IgG. Three points were reduced in the IAIHG scoring system based on histological findings typical of NASH as described above.
Simplified AIH criteria were also applied in the current cohort [18]. The Paris criteria for diagnosis of overlap syndrome was not adopted in the current study [19].
Biochemical remission was defined as normal ALT levels at least 1 year after steroidal agents were first prescribed [1]. Moderate to severe AIH was defined using AST > × 5 ULN, or IgG > 2 ULN or confluent necrosis in liver pathology referring to a past study [20].
Statistical analyses
Continuous variables were presented as median and interquartile range and were analyzed using the Mann-Whitney U test or Spearman’s rank correlation coefficient. Categorical variables were analyzed using Fisher’s exact test or Chi-squared test depending on patient number in each category. Statistical analyses mentioned above were performed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA). Cox proportional hazard model was analyzed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), a graphical user interface for R software (The R Foundation for Statistical Computing, Vienna, Austria) [21, 22]. P < 0.05 was considered statistically significant.