Prognosis of Probable Autoimmune Hepatitis Patients in a Retrospective Cohort Study

Background: Autoimmune hepatitis (AIH) is an idiopathic inammatory liver disease with genetic susceptibility and unknown environmental triggers, which results in failures of physiologic immunotolerance and destruction of the liver tissues. The gold standard for diagnosis is the International Autoimmune Hepatitis Group (IAIHG) scoring system: the disease is classied as denite or probable according to the scores. However, conventional research on probable AIH has focused on the Caucasian population and there is little data pertaining to the Asian population. Therefore, this study aimed to assess and compare the prognosis of Japanese patients with probable and denite AIH. Methods: Patients with probable and denite AIH diagnosed based on IAIHG scores between 1987 and 2018 were enrolled in this retrospective study. Results: Seventy-two patients with denite AIH and 49 patients with probable AIH were evaluated in the study. Univariate analysis revealed age, brosis stage 4, and the brosis-4 index were prognostic factors for overall survival. Multivariate analysis indicated that age and liver cirrhosis signicantly affected the overall survival. When the cut off albumin-bilirubin score was set appropriately, cirrhosis was differentially diagnosed using albumin-bilirubin score with 100% sensitivity and 70.5% specicity. Classication of probable or denite disease did not alter overall survival with statistical signicance. Conclusions: Our ndings suggest that probable AIH should be managed as denite AIH is managed in Japanese population. The albumin-bilirubin score helps identify liver cirrhosis and is a prognostic biomarker for overall survival.


Background
Autoimmune hepatitis is an idiopathic in ammatory liver disease based on genetic susceptibility and unknown environmental triggers, which result in failures of physiologic immunotolerance and destruction of liver tissues [1][2][3]. The clinical presentation is characterized by a variety of aspects, including an increase in serum hepatobiliary enzyme, serum IgG, elevation of autoantibody titer, histopathological ndings of liver biopsy specimens, and complications of other autoimmune diseases [4][5][6].
The gold standard for diagnosis is the International Autoimmune Hepatitis Group (IAIHG) scoring system, which classi es the disease as de nite or probable according to diagnostic scores [7]. The epidemiology, baseline characteristics, and treatment response of de nite autoimmune hepatitis (AIH) are already established [8]. The differences in epidemiology between Caucasian and Asian AIH have also been summarized in detail [9,10]. However, conventional research on probable AIH has focused on the Caucasian population and there is far less available evidence on the Asian population. The aim of this study was to clarify the prognosis of probable AIH patients in a Japanese population compared to the de nite disease.

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 de nite 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. De nite and probable AIH patients were analyzed further.

Histopathological analysis
Pathological ndings 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 de nite 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 de nite 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 ndings typical of NASH as described above.
Simpli ed 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 de ned as normal ALT levels at least 1 year after steroidal agents were rst prescribed [1]. Moderate to severe AIH was de ned using AST > × 5 ULN, or IgG > 2 ULN or con uent 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 coe cient. 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 signi cant.

Characteristics of the patients
The clinical data of 121 patients, including liver histopathology, were reviewed in the current study ( Table  1). The cohort consisted of 72 patients with de nite AIH (the de nite cohort) and 49 patients with probable AIH (the probable cohort). Compared to de nite AIH patients, probable patients included a higher proportion of males. They were more likely to be negative for ANA and positive for AMA or antimitochondrial M2 antibody and they had higher albumin levels, lower IgG levels, and lower ALBI scores.
No patients were positive for LKM-1 in either cohort. Two patients of de nite AIH presented positive for anti-smooth muscle antibody, but it did not contribute to diagnostic points because the patients were also positive for ANA. More patients in the probable cohort were positive for human leukocyte antigen death receptor 4 (HLA-DR4) compared to the de nite cohort.
In the probable cohort, chronic nonsuppurative destructive cholangitis was detected in three patients and granuloma was detected in one patient according to histopathological examinations, which show typical PBC features. One patient presented with hepatocyte ballooning with background steatosis equal to or more than 5%, which is the prominent histopathological feature of NASH. Positive hepatitis B antigens were detected in two patients, HCV-RNA was detected in ve patients, and one patient had a history of taking pregabalin [23,24]. Two patients showed remarkable alcohol consumption.
Emperipolesis was pointed out in the pathological work up in two patients of de nite disease. Compared to the probable AIH patients, no patients in the de nite AIH cohort showed the prominent features of NASH, HBV, or HCV infection, or history of alcohol or drug use.
In the total cohort, IAIHG scores were correlated with the simpli ed IAIHG scores with an r value of 0.6252 (95% con dence interval, 0.4988 -0.7255; p value < 0.0001). Seventy-nine (65.3 %) of total 121 patients were credited with the simpli ed score equal to or above 6 points.
Follow up of the patients Patients were followed up for a median of 4 years with the longest follow up period of 30 years (Table 1). Steroidal agents were administered to 41.2% of patients in the de nite cohort. The number of patients in the probable cohort who underwent steroidal therapy was signi cantly smaller compared to the de nite cohort. Two patients in the de nite cohort were diagnosed with HCC during the follow up period but survived through the observation. In this study, 12.2% of probable AIH patients and 5.6% of de nite patients died, and this difference was not signi cant. The cause of death was identi ed in seven patients -three due to liver failure, one due to rupture of abdominal aortic aneurysm, one due to bacterial peritonitis, one due to interstitial pneumonia, and one due to cryptococcus meningitis.

Overall survival
Univariate analysis was performed to determine the predictive prognostic factors for overall survival in the total cohort. As shown in Table 2, age, brosis stage 4, and the brosis-4 index were prognostic factors for survival. The hazard ratio of 1.885 for probable disease over de nite one resulted in no statistical signi cance.
Multivariate analysis was performed to determine independent risk factors for overall survival employing ve parameters: de nite or probable disease, age, moderate to severe disease, Fibrosis stage 4, and nonremission under steroidal therapy (Table 2). Sex was not included in the analysis because sex was incorporated in IAIHG scoring system. The result indicated that age and brosis stage 4 signi cantly contributed to the overall survival. The hazard ratio for age was 1.099 per year in the multivariate analysis while the hazard ratio for brosis stage 4 was 8.943. Although the hazard ratio for probable disease increased up to 2.446, it remained statistically non-signi cant.
Surrogate marker for brosis stage 4 Fibrosis stage 4 was identi ed as a prognostic factor for overall survival in the current cohort. As surrogate biomarkers for brosis stage 4, brosis-4 index and ALBI score were evaluated. As shown in Figure 1a, the median brosis-4 index values did not differ between patients with brosis stage 1 to 3 and patients with stage 4 (p = 0.2225), but ALBI scores were different in patients with brosis stage 4 compared to those with stage 1 to 3 (p = 0.0001), as indicated in Figure 1b. Receiver operating characteristic (ROC) analysis revealed that the area under curve for ALBI score was 0.8586. When the cut off ALBI score value was set at -1.994, brosis stage 4 was differentially diagnosed using ALBI score with 100% of sensitivity and 70.5% of speci city (p = 0.0004, Figure 1c).

Discussion
The current study aimed to clarify prognosis of probable AIH patients in the Japanese population. The data revealed that 1) prognosis of probable AIH does not differ from that of de nite disease, 2) greater age and brosis stage 4 at baseline are independent risk factors for overall survival, and 3) the ALBI score serves as a surrogate marker for brosis stage 4 in the Japanese population.
A couple of studies have reported prognosis of probable autoimmune hepatitis. In an English population, 78 probable AIH patients were compared to 167 de nite patients, and no signi cant difference in liverrelated death or transplant-free survival [25]. Summing up probable and de nite disease patients, cirrhosis at diagnosis and biochemical remission were indicated as prognostic factors in that study.
Czaja et al. also performed a retrospective study of 17 probable AIH patients de ned using the IAIHG scoring system [26]. They concluded that the classi cation of de nite and probable was based mainly on sex and the alteration of laboratory data, which do not re ect disease severity or treatment response. The British Society of Gastroenterology stated in a public announcement that probable and de nite AIH appeared to be the same disease in terms of outlook and response to immunosuppression [1]. A largesized cross-sectional study reported in the United Kingdoms in 2018 omitted comparative analyses between the probable and de nite AIH patients [27]. Another study prospectively evaluated prognostic factors of AIH using a cohort of 325 patients including both probable and de nite diseases [28]. The study identi ed age at diagnosis as a risk factor for liver-related death or transplant-free survival.
In the current study, classi cation of probable or de nite disease did not stratify patients according to biochemical response rate and overall survival, similar to past reports. Fibrosis stage 4 and age were also extracted as prognostic factors negatively correlated with overall survival in the total cohort. Our data showed that prognosis of probable AIH was not different to that of de nite AIH in the Japanese population, orchestrating the results in Caucasian cohorts. Correlation of IAIHG and the simpli ed criteria was also comparable with that of a past study [27], as 2/3 of AIH patients diagnosed using IAIHG met the simpli ed criteria in the current data.
However, one apparent difference should be noted. Immunosuppression therapy was applied to 224 of 245 patients (91.4%) in the past report [25], while steroidal agents were prescribed for 41 of 121 patients (33.9%) alone in the current cohort. Application of immunosuppression therapy is generally considered in patients with an AST > × 5 ULN, or γ-globulin > 2 ULN or con uent necrosis in liver pathology at diagnosis [1,20]. A Japanese nationwide survey using questionnaires reported that AIH patients presented with a median AST of 177 U/l based on a total of 1668 AIH patients [29]. Contrary to this, a past study in England reported a median AST 489 U/l for de nite AIH and 235 U/l for probable AIH at baseline [25]. It can be said that AIH disease severity is milder in the Japanese population than in English populations. In the current study, Japanese AIH patients presented with further milder in ammatory activity on liver biopsy examination.
In the Asian AIH population, ANA and HLA-DR4 results are more frequently positive compared to the Caucasian population. In the current study, 95.8% of de nite AIH patients were positive for ANA and 90.0% of patients were positive in the total cohort, orchestrating the results of the Japanese nationwide survey [29]. However, HLA-DR4 was positive in a little 19.0% of the total cohort in the current study. Furthermore, probable AIH patients were positive for HLA-DR4 with larger a proportion than de nite AIH ones. This discrepancy may be due to two reasons. Firstly, if a patient is positive for ANA, HLA-DR4 does not contribute to diagnosis of AIH in the IAIHG scoring system [7] and about 90% of Japanese AIH patients are positive for ANA [29] and secondly, HLA typing is not covered by insurance in Japan. Thus, clinicians should omit to determine HLA typing in clinical practice.
Anti-soluble liver antigen (SLA) antibody has not been covered by public insurance in Japan, either. Anti-SLA antibody is frequently detected in patients negative for conventional autoantibodies [30,31]. The antibody is reported to correlate higher relapse rate of the disease with poorer prognosis [32]. However, no data were available for anti-SLA antibody in our study. The IAIHG scoring system does not speci cally mention how to exclude NASH from the diagnosis of AIH, or how to incorporate the typical features of NASH into the scoring system [7]. The cross-sectional study performed in the UK adopted a criterion for the issue as more than mild steatosis regardless of hepatocyte ballooning [27]. A past report presented evidence that baseline liver status was more frequently cirrhotic, and that patient survival was worse in AIH with NASH than pure AIH [33]. In the current study, hepatic steatosis equal to or more than 5% accompanied by hepatocyte ballooning was de ned as NASH [17]. The complications of NASH with AIH was not subject to statistical analysis in our cohort because only a single patient tted the above criteria.
Fibrosis stage 4 contributed to overall survival in the current study. In the management of AIH after initial diagnosis, surrogate markers for brosis stage should be useful as an alternative to liver biopsy examination. Liver biopsy maneuvers sometimes complicate liver hemorrhage. Our data clari ed that the ALBI score had better diagnostic potential for distinguishing brosis stage 4 from stage 1 to 3 compared to the conventional biomarker of liver brosis, the brosis-4 index. This may be partially explained given that the value of brosis-4 index does not re ect the actual brosis stage because of the uctuation of AST and ALT values depending on the in ammatory activity of the disease.
The ALBI score was originally established to classify cirrhosis patients according to their prognosis [15].
An ALBI score less than -2.600 means better prognosis and is classi ed grade 1, an ALBI score between -2.600 and less than -1.390 is means moderate prognosis and is classi ed grade 2, an ALBI score more than -1.390 means a worse prognosis and is classi ed as grade 3. Currently, application of ALBI score is limited in cirrhosis patients, similar to the Child Pugh score. However, our previous studies have clari ed that the ALBI score has the potential to differentially diagnose brosis stage 4 from stage 1 to 3 in HCV infection, HBV infection, and PBC [34][35][36]. In AIH, ALBI score was able to diagnose brosis stage 4 regardless of the in ammatory severity of the hepatitis at any stage of the disease.

Conclusions
In the Japanese cohort, we have shown that: 1) the classi cation of AIH, probable or de nite, does not alter overall survival, 2) greater age and brosis stage 4 at baseline independently strati ed overall survival, similar to the results for the Caucasian population, and 3) brosis stage 4 was diagnosed more accurately using ALBI scores compared to the conventional biomarker, brosis-4 index. De nite and probable AIH patients in Japan can be monitored their liver brosis progression using ALBI score.

Abbreviations
Anti-mitochondrial antibody, AMA; anti-nuclear antibody, ANA; autoimmune hepatitis, AIH; International autoimmune hepatitis group , IAIHG; Nonalcoholic steatohepatitis, NASH; Primary biliary cholangitis, PBC; Receiver operating characteristic, ROC; smooth muscle antibody, SMA; type-1 liver-kidney microsomal antibody, LKM-1, 95% con dence interval, 95% CI Declarations Ethics approval and consent to participate Participants in the study provided informed consent, and the study design was approved by the appropriate ethics review board.

Consent for publication
Not applicable.

Availability of data and materials
Our data will be available from the corresponding author on reasonable request.

Competing interests
No authors have any con ict of interest.

Funding
This study received no nancial support.

Authors' contributions
The author's contributions are as follows. The study concept and design were con gured by KF, data were acquired by KO, TK, and JT, data were analyzed and interpreted by AM, critical revision was performed by HK, KT and TH, and the study was supervised by TM. IAIHG, International Autoimmune Hepatitis Group; T-Bil, total bilirubin; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; IgG, immunoglobulin G; Plt, platelet; ANA, antinuclear antibody; ALBI, albumin-bilirubin score; CI, con dence interval Figures Figure 1 Staging of liver brosis in autoimmune hepatitis a) using albumin bilirubin score and b) using the brosis-4 index. The albumin-bilirubin score differentiates stage 4 from stage 1-3 (p = 0.0001), but the brosis-4 index did not (p = 0.2225). c) Receiver operating characteristic analysis reveals that an albumin-