KASL clinical practice guidelines for management of autoimmune hepatitis 2022

AIH-PBC overlap syndrome

The “Paris criteria” is the most common and effective method to diagnose the AIH-PBC overlap syndrome [105]. It requires at least two of the following three diagnostic criteria for each disease. Two of the following three criteria for PBC should be met: (1) serum ALP level ≥2-fold the upper limit of normal (ULN) range or serum gamma-glutamyl transferase (GGT) level ≥5-fold ULN; (2) presence of AMA; and (3) a liver biopsy specimen showing florid bile duct lesions (non-suppurative destructive cholangitis of interlobular bile duct). For AIH, it requires two of the following three diagnostic criteria: (1) ALT ≥5-fold ULN; (2) serum IgG level ≥2-fold ULN or presence of SMA; and (3) a liver biopsy with moderate or severe interface hepatitis [106]. Since the revised diagnostic scoring system excludes PBC, a Korean study was performed to demonstrate that the simplified diagnostic scoring system can help diagnose overlap syndrome; however, due to the small sample size of the study, it had limited clinical application [107].

AIH-PSC overlap syndrome

Criteria for the diagnosis of AIH-PSC overlap syndrome include the presence of typical features of AIH, absence of AMA, and evidence of large duct PSC based on bead-like appearance characterized by focal narrowing and dilatation of the bile duct on magnetic resonance cholangiopancreatography (MRCP) or endoscopic retrograde cholangiography (ERCP), or evidence of small-duct PSC based on characteristic fibrous obliterative cholangitis on histology [108]. In a study of patients with PSC, it was reported that the revised diagnostic scoring system could help diagnose AIH-PSC overlap syndrome [109].

AIH-like drug-induced liver injury

Clinical manifestations and pathological findings of DILI caused by an unpredictable idiosyncratic drug reaction or hypersensitive drug reaction were similar to those of AIH in about 2-17% of patients reported as AIH [110-112]. The leading causative agents are known to be nitrofurantoin, minocycline, alpha-methyl DOPA, and hydralazine [113]. DILI can resemble the clinical manifestations of AIH by causing the formation of serum autoantibodies and gammaglobulinemia. A liver biopsy can be performed to differentiate between DILI and AIH. However, DILI may be difficult to distinguish from AIH due to the manifestation of interface hepatitis and plasma cell infiltration [114]. Therefore, drugs and supplements exposed before disease onset should be clearly identified. After drug exposure, the latency period of AIH-like DILI varies greatly, ranging from 1–8 weeks to 3–12 months [21,115]. Moreover, the assessment of response to and recurrence after glucocorticoid therapy is helpful in differentiating between AIH and DILI [113].

AIH-like DILI mainly affects women, and acute hepatitis is the common manifestation of DILI in more than 60% of all cases. Cardinal symptoms include nausea, vomiting, lethargy, and right upper quadrant pain. Approximately 30% of DILI patients display signs of drug hypersensitivity reactions, such as fever, rash, and increased eosinophils [21]. In genetic tests, HLA DRB1*03:01 or HLA DRB1*04:01 are found to be similar to healthy controls, and autoimmune disease is rarely accompanied [116]. When a previous study reviewed 261 patients diagnosed with AIH over a decade, AIH-like DILI was detected in 24 patients, accounting for 9.2% of all cases [110]. The median age was 53 years (interquartile range, 24–61), and nitrofurantoin and minocycline were the most common agents associated with DILI [110]. Liver enzyme levels were elevated up to 5–20 times the normal amount, while ALP increased slightly. Serum bilirubin varied to over 20 mg/dL from the normal range, and elevated gamma globulin levels, ANA positivity (83%), and SMA positivity (50%) were observed [110]. Furthermore, the symptoms were mitigated by stopping the causative drug and receiving glucocorticoid treatment. No relapse was observed after discontinuation of immunosuppressive treatment, and none of the patients progressed to LC [110]. Immune-related adverse events are being increasingly reported, with the growing use of immune checkpoint inhibitors that activates immune cells to block various cancers. DILI caused by immune checkpoint inhibitors is highly responsive to glucocorticoid therapy, and it usually shows negative or low levels of serum ANA and SMA and has normal gamma globulin levels [117].

Monitoring is warranted to check for improvement in clinical manifestations and laboratory findings without recurrence after stopping the causative agents. Most cases of DILI improve within a month, but may rarely persist for more than 3 months. According to Hy’s Law criteria, when serum AST and ALT levels are elevated more than three times the ULN and serum bilirubin is greater than two times the ULN, it leads to the risk of death or liver transplantation in approximately 9–12% of cases [118,119]. The use of glucocorticoids is considered when symptoms show no improvement despite the suspension of drug use and meet Hy’s Law criteria. The diagnosis of DILI can be confirmed when normal values are maintained in routine blood tests after the withdrawal of glucocorticoid therapy. On the contrary, repeatedly elevated liver enzymes may indicate AIH. Relapsing hepatitis is managed the same way as AIH using immunosuppressive agents [120]. Most patients with AIH-like DILI have a good prognosis, but this condition may rarely result in death, in about 5%, due to idiosyncratic drug response and require liver transplantation in about 4.5% of cases [121].

Serum biomarkers

The degree of liver fibrosis of AIH patients can be estimated using serum panel. The serum panel including FibroTest which combines five serum biochemical markers (α-macroglobulin, apolipoprotein A1, haptoglobin, total bilirubin, GGT) with patient age and gender [122-124], aspartate aminotransferase-to-platelet ratio index (APRI) [125], Fibrosis-4 index (FIB-4) which combines patient age with measurements of 3 biomarkers (AST, ALT and platelet) [126,127], enhanced liver fibrosis test (ELF) which combines tissue inhibitor of metalloproteinases 1 (TIMP-1), amino-terminal propeptide of type III procollagen (PIIINP) and hyaluronic acid (HA) [128,129], are well-validated non-invasive tests in viral hepatitis. However, their role in predicting the progression of liver fibrosis, long-term prognosis, and development of hepatocellular carcinoma (HCC) in AIH remains unknown [130]. Another serum marker Mac-2-binding protein glycan isomer (M2BPGi) has been studied in Japanese AIH patients, but its applicability in Korean patients requires further validation [131,132].

Imaging modalities

The ultrasound-based measurements of liver stiffness comprise transient elastography (TE), 2D shear wave elastography (SWE), and point SWE, while other methods include magnetic resonance elastography (MRE).

In AIH, studies based on FibroScan^® are relatively more common than other non-invasive imaging modalities [133-136]. The median value of liver stiffness measurement (LSM) in AIH patients was higher than that of healthy controls (11.2±8.2 kPa vs. 4.3±1.4 kPa, P<0.01) [136]. A robust positive correlation was observed between LSM and histological fibrosis stage [134]. However, the LSM value was higher within 3 months of treatment, and the area under the receiver operating characteristic curve (AUROC) at 6 months was higher than that at 3 months of treatment (≥F2, 0.68 vs. 0.97; ≥F3, 0.8 vs. 1.0; F4, 0.71 vs. 1.0). The best cut-off values for ≥F2, ≥F3, and F4 at 6 months of treatment were 5.8 kPa, 10.5 kPa, and 16.0 kPa, respectively [134]. Since hepatic inflammation impacts LSM, ≥F3 can be more accurately diagnosed after 6 months of treatment when hepatic inflammation is resolved [134]. In another retrospective study, patients who failed to achieve complete biochemical remission showed a slight increase in LSM (+1.7%/year; 95% CI, -6.0% to 12.1%; P=0.19), while a significant decrease in LSM (-7.5%/year; 95% CI, -11% to -2.0%; P=0.0003) was observed in the complete biochemical remission group, indicating that fibrosis regression can be monitored by TE [135]. According to a meta-analysis, TE performed better than serum markers, FIB-4 and APRI, in staging advanced fibrosis ≥F3 [130,137].

The AUROCs for 2D-SWE in diagnosing ≥F2, ≥F3, and F4 were 0.85, 0.85, and 0.86, respectively [138]. Other studies reported similar AUROCs ranging from 0.781 to 0.84 [139,140], showing higher predictive efficacy compared to APRI and FIB-4 (AUROC, 0.84 vs. 0.57 vs. 0.63) [140]. A single-center study in South Korea on point SWE which generates shear wave in one area revealed an AUROC of 0.8, similar to the results of studies from other countries that showed point SWE outperforming APRI and FIB-4 [141].

The MRE has the advantage of evaluating the whole liver, and MRE appeared to outperform TE for staging hepatic fibrosis in some studies focusing on other liver diseases [142]. One study based on 36 patients showed that the AUROCs for advanced fibrosis (≥F3) and cirrhosis (F4) were 0.97 and 0.98, respectively [143]. The best cut-off values for ≥F3 and F4 were 4.1 kPa (sensitivity 89.5%; specificity 100%) and 4.5 kPa (sensitivity 92.5%; specificity 96%), respectively, revealing a very high diagnostic accuracy [143]. Although a study comparing MRE with TE is lacking, MRE outperformed APRI and FIB-4; and therefore, further evaluation on the role of MRE in AIH patients is required in the future.

Treatment aims and definitions of treatment endpoints

The goals of AIH treatment are to minimize the risk of complications caused by drugs, control the liver inflammation, and achieve remission to suppress the progression of liver disease. To achieve these aims, long-term or permanent maintenance therapy after remission is required in most patients with AIH.

The ideal biochemical response is the normalization of serum ALT, AST, and IgG, and the ideal treatment response is the loss of histologic inflammation along with the biochemical response [21,93,144-148]. Even with the biochemical response, histologic inflammation often persists. Since aminotransferases and IgG do not reflect the activity of histologic inflammation, especially in the case of cirrhosis [82,149], liver biopsy may be necessary to confirm the loss of histologic inflammation. A study with paired liver biopsy in 120 patients with a biochemical response for more than 6 months showed that 46% of patients had a histological activity with an Ishak score of 4 or higher, and it was an independent factor associated with death or liver transplantation [149].

Complete biochemical response is the normalization of serum transaminases and IgG below the ULN within 6 months of treatment (Table 5) [150]. Insufficient response is the lack of complete biochemical response, determined no later than 6 months after the initiation of treatment. Non-response is <50% decrease of serum transaminases within 4 weeks after the initiation of treatment. Remission refers to a case where the hepatitis activity index (HAI) of liver tissue is less than 4 out of 18 points [150]. Intolerance to treatment is any adverse event possibly related to treatment as assessed by the treating physician, leading to potential discontinuation of the drug [21].

Table 5.

Endpoints for AIH treatment as proposed by the International Autoimmune Hepatitis Working Group after a consensus process

Persistent elevation of AST or ALT level during treatment is known to predict the progression of liver diseases and poor prognosis, such as recurrence, histological activity, cirrhosis, and hepatocellular carcinoma, after treatment is discontinued [146-148,151,152]. According to a retrospective study of 132 patients with AIH, patients whose serum biochemical indicators did not return to normal had a 3–11 times higher risk of relapse after discontinuation of treatment compared to patients with normal serum biochemistry [146]. Another study reported that only 4% of patients with normal serum biochemical indicators experienced histological deterioration, while 54.5% of patients without normalization experienced histological and clinical deterioration [151]. In addition, since serological indicators, especially ALT and IgG, are closely related to histological activity [153], normalization of these can be used as indicators to predict histological remission.

Treatment indications

Active research on the treatment of AIH was conducted from the 1960s to the 1970s [154-158], and treatment regimens based on the results of these studies are valid to date. Since the hepatitis C virus was discovered in 1989, there is a possibility that chronic hepatitis C patients may have been included among patients diagnosed with AIH before 1989, and therefore, some hepatitis C patients may have been included in the initial clinical trial. In a prospective randomized controlled study conducted for the first time in patients with chronic active hepatitis, the placebo group without treatment showed a high mortality rate of 56% at the 72-month follow-up, whereas the mortality rate of patients treated with prednisolone decreased to 14% [154]. In several subsequent randomized controlled studies, patients with chronic active hepatitis who were not treated showed a high mortality rate (41% at 3–3.5 years of follow-up), whereas those treated with prednisone alone or prednisone plus AZA showed a reduced mortality by 6–10% [155,157]. Accordingly, it has been confirmed that untreated active AIH has a very poor prognosis, and that appropriate immunosuppressive therapy improves liver function and increases survival.

Evidence on the natural course and benefits of immunosuppressive treatment in asymptomatic AIH patients with mild inflammatory activity is still insufficient. A Canadian single-center cohort study of 126 patients with AIH reported a 10-year survival rate of 80.0% (95% CI, 62.5–97.5%) for patients with asymptomatic AIH, and untreated asymptomatic patients showed a statistically insignificant survival difference compared to asymptomatic patients who received treatment [22]. On the other hand, in another retrospective study conducted in the United States, some asymptomatic patients with mild activity reached remission without treatment, but the rate of reaching remission was significantly lower than that of patients who received immunosuppressive therapy (12% vs. 63%, P= 0.006), and their 10-year survival rate was also significantly lower (67% vs. 98%, P<0.01) [159]. AIH may have reached remission spontaneously without treatment, but the spontaneous remission did not persist after recurrence [22,159]. In a large retrospective study of 305 patients with AIH, asymptomatic patients had significantly lower biochemical and histological activity compared to symptomatic patients, but the response rate to immunosuppressive therapy (complete response rate P=0.558; non-response rate P=0.462) and liver-related prognosis (P=0.975) were found to be similar between the two groups [24]. If AIH is not treated, it is difficult to predict the disease course as the activity continuously changes; and a significant number of asymptomatic patients develop symptoms (25.8–69.6%) [22,160], experience liver disease progression (22.2–50%) [24,159,160], or progress to hepatocellular carcinoma, end-stage liver disease, or liver failure [159,160]. In a multicenter longitudinal cohort study in the UK, all patients with AIH treated with glucocorticoids had lower overall mortality and lower liver transplant rates compared to untreated patients (hazard ratio [HR], 0.25; 95% CI; 0.14–0.45; P<0.001); and in particular, the overall mortality and liver transplantation rates were significantly lower (HR, 0.13; 95% CI, 0.04–0.42; P=0.001) when even asymptomatic patients were treated [161].

Considering the natural course of AIH and the effect of immunosuppressive therapy, active AIH patients with abnormal clinical and laboratory findings (elevation of AST, ALT, and IgG) or liver tissue findings suggesting intrahepatic inflammation (HAI ≥4) are subject to immunosuppressive treatment. When treatment is withheld in asymptomatic inactive patients with an HAI score of less than 4 without advanced fibrosis, liver enzyme levels and IgG markers should be monitored regularly.

First-line standard therapy

For the induction of remission of AIH, prednisolone 20–40 mg and AZA 50–150 mg are administered in combination daily, or prednisolone 40–60 mg alone daily (Fig. 5). Combination therapy of prednisolone at higher doses (up to 1 mg/kg/day) and AZA can induce rapid remission in patients with AIH without cirrhosis [162], but steroid-related side effects should be taken into consideration.

Figure 5.

Induction strategy for autoimmune hepatitis. AIH, autoimmune hepatitis; AZA, azathioprine; DILI, drug-induced liver injury; HAI, hepatitis activity index; NUDT15, Nudix hydrolase15; TPMT, thiopurine S-methyltransferase. ^*Delayed institution of AZA by 2 weeks can be considered. ^†Emergent evaluation for liver transplantation should be considered for patients with ALF.

The efficacy of prednisolone alone or AZA combination therapy in AIH has been demonstrated through several randomized controlled trials [154-158]. A systematic review of these randomized controlled trials showed similar remission rates between predniso(lo)ne monotherapy and AZA combination therapy (42% vs. 43%; relative risk [RR], 0.98; 95% CI, 0.65–1.47), and fewer drug-related adverse events occurred with AZA combination therapy [163]. Prednisolone and AZA combination therapy is similar in efficacy to prednisolone monotherapy, but has advantages in terms of adverse events and is preferred as the first-line treatment [20]. On the other hand, when the treatment duration is expected to be shorter than 6 months, such as AIH-like DILI, or when AZA is contraindicated, prednisolone monotherapy is recommended [21].

According to a retrospective study on the initial dose of prednisolone, when comparing prednisolone 30 mg and 40 mg as a combination therapy with AZA, the remission rate at 3 months of treatment was higher in the 40 mg group (69.2% vs. 43.8%, P=0.031), but there was no statistically significant difference in remission rates at 6 months and 12 months (79.5% vs. 59.4%, P=0.065; 89.5% vs. 80.6%, P=0.30) and in recurrence rate during maintenance therapy (35.9% vs. 50%, P=0.23) [164]. In a multicenter retrospective study conducted in Europe, there was no significant difference in biochemical response rates at 6 months between the high-dose prednisolone (≥0.5 mg/kg/day) and low-dose prednisolone (<0.5 mg/kg/day) treated groups (70.5% vs. 64.7%, P=0.61). There was no statistically significant difference in glucocorticoid-related adverse events between the two groups, but the incidence of glucocorticoid-induced diabetes (7.7% vs. 3.9%, P=0.13) and osteoporosis (6.4% vs. 2.6%, P=0.09) was higher in the high-dose administration group [165]. A meta-analysis of 25 studies on glucocorticoid doses reported that high-dose glucocorticoid (60 mg/day or 1 mg/kg/day) administration had a higher biochemical remission rate (79% vs. 72%) compared to low-dose glucocorticoids (40–50 mg/day or 0.5 mg/kg/day), but liver transplantation or mortality (3% vs. 1%) and glucocorticoid-related adverse events (42% vs. 39%) were also higher [166].

In combination therapy for inducing remission of AIH, AZA can be administered simultaneously with prednisolone or administered sequentially with an interval of about 2 weeks. When administered sequentially, the response to prednisolone can be evaluated with prednisolone administration alone during the first 2 weeks of treatment, eliminating the uncertainty of diagnosis and accurately evaluating the treatment response by excluding AZA-induced hepatotoxicity that may rarely occur in severe liver disease [20,167]. In addition, the risk of complications can be predicted by evaluating the patient’s NUDT15 (Nudix hydrolase 15) and TPMT (thiopurine S-methyltransferase) gene mutations during the first 2 weeks of not administering AZA [21]. AZA is initially administered in combination with prednisolone at a dose of 50 mg/day and may be increased to 150 mg/day or 2 mg/kg/day depending on toxicity and response to treatment. If a NUDT15 or TPMT gene variant is present, AZA metabolism is impaired and the risk of cytopenia due to bone marrow suppression increases. Since patients homozygous for any risk variants have almost no enzymatic activity, prednisolone monotherapy or alternative therapy without AZA should be considered. Even the patients who are heterozygous for risk variants can also have a risk of bone marrow suppression, so the dose of AZA should be reduced. When considering a dose of 2 mg/kg/day or more, the starting dose of AZA should be reduced by 30–80% and adjusted based on the degree of myelosuppression [168]. In particular, more attention should be paid to patients with risk variants in both the NUDT15 and TPMT genotypes.

If there is a biochemical response with an initial dose of prednisolone and AZA, monitoring should be conducted every 1–2 weeks and prednisolone gradually reduced to a dose that maintains the biochemical response or 20 mg/day while maintaining AZA. Then, monitoring should be performed every 2–4 weeks and prednisolone gradually reduced by 2.5–5 mg to maintain 5–10 mg/day or a dose that maintains the biochemical response. After achieving a complete biochemical response, prednisolone can be discontinued while maintaining AZA. In several randomized controlled trials of maintenance therapy [169-171], AZA alone maintenance therapy showed a higher sustained remission rate than predniso(lo)ne alone maintenance therapy (92% vs. 68%; RR, 1.31; 95% CI, 1.07–1.70), and there was no significant difference in the sustained remission rate compared to predniso(lo)ne plus AZA maintenance therapy (92% vs. 96%; RR, 1.06; 95% CI, 0.94–1.20) [163]. In addition, high-dose AZA monotherapy (2 mg/kg/day) reduced glucocorticoid-induced adverse events and recurrence, showing a high remission persistence rate of 83% for an average of 67 months [170,172]. If leukopenia or thrombocytopenia occurs during AZA treatment, the dose should be reduced or discontinued; especially, if cytopenia does not recover within 1–2 weeks, AZA should be discontinued. Care should be taken in patients with LC due to the high incidence of AZA-induced cytopenia [173,174]. When AZA administration is impossible due to adverse events, the lowest dose of prednisolone alone can be administered as maintenance therapy. However, long-term administration of prednisolone in doses exceeding 10 mg per day may cause frequent steroid-related adverse events. Therefore, it is recommended to administer the lowest dose of prednisolone that maintains the biochemical reaction, keeping the dose below 10 mg/day if possible [175].

Alternative first-line therapy

According to a prospective randomized controlled study on the efficacy and safety of budesonide and AZA combination therapy, budesonide (9 mg/day) and AZA (1–2 mg/kg/day) combination therapy showed a significantly higher 6-month biochemical remission rate (60% vs. 38.8%, P=0.001) and lower side effects of glucocorticoids (26% vs. 51.5%, P<0.001) compared to prednisolone (40 mg/day) and AZA (1–2 mg/kg/day) combination therapy in patients with AIH without cirrhosis [176]. Since budesonide has a high (>90%) first-pass effect in the liver, it has fewer adverse effects caused by glucocorticoids and can be advantageous in terms of bone density in the long term [177-179]. However, budesonide bypasses the liver due to portal systemic shunt in patients with LC, which reduces the efficacy of glucocorticoids and increases glucocorticoid-induced adverse effects [180,181]. Therefore, as a first-line treatment, budesonide and AZA combination therapy can be selectively considered in patients with AIH without cirrhosis who are highly likely to have glucocorticoid-induced adverse effects, or when prednisolone administration is not possible due to adverse effects. However, as of 2022, oral budesonide cannot be used in South Korea as it is not commercially available.

If the administration of AZA is not possible, mycophenolate mofetil (MMF) can be used as an alternative first-line treatment. A single-center prospective study reported a remission rate of 71.6% when MMF (1.5–2 g/day) was administered in combination with prednisolone as the first-line treatment for AIH, and 78.2% of them maintained remission with MMF-only (1–1.5 g/day) maintenance therapy [182]. As a result of a meta-analysis of seven prospective and retrospective studies, the predniso(lo)ne-MMF combination therapy showed significantly higher normalization rates of AST, ALT (OR, 1.49; 95% CI, 1.02–2.18), and IgG (OR, 1.87; 95% CI, 1.21–2.88), and significantly lower non-response rates (OR, 0.55; 95% CI, 0.36–0.85) compared to the predniso(lo)ne-AZA combination therapy [183]. Although MMF has been proven to be effective when administered in combination with prednisolone as a first-line treatment for AIH, it is recommended as an alternative treatment to AZA due to insufficient studies to date.

Acute severe AIH and acute liver failure due to AIH

In acute severe AIH or ALF due to AIH, the efficacy and optimal dose of glucocorticoid treatment have not yet been clearly proven due to the high rate of treatment failure [184], the possibility of delayed liver transplantation due to initial medical treatment [185,186], and the risk of infection due to glucocorticoid administration [187].

According to a study of 72 patients with acute severe AIH with jaundice, the treatment failure rate was as high as 18% when treated with predniso(lo)ne at a dose of 40 to 60 mg/day. The higher the serum level of bilirubin and PT, the higher the risk of treatment failure, and the higher the risk of death and emergency liver transplantation in case of treatment failure [184]. In a French single-center retrospective study of 16 patients with acute severe AIH or ALF due to AIH (63% of whom had hepatic encephalopathy, median PT INR of 5.4), 12 patients received corticosteroid therapy, 11 patients underwent liver transplantation, one patient died, and three pateints developed severe sepsis, reporting that corticosteroid treatment increased the incidence of infection without improving the prognosis [187]. Meanwhile, in a study of 32 patients with acute severe AIH without hepatic encephalopathy, all of the untreated patients required emergency liver transplantation, whereas only 43% of patients treated with corticosteroids (mean dose of 40 mg/day of predniso(lo)ne) received emergency liver transplantation (P=0.004), and the sepsis incidence and mortality were not significantly different between the two groups (11% vs. 26%, P=0.6; 22% vs. 17%, P=0.99) [188]. Another study in patients with acute severe AIH also reported favorable long-term survival (97% during 5.3 years) without progression to liver failure or liver transplantation with early administration of high-dose glucocorticoids (1.5 mg/kg/day of prednisolone) [189].

Collectively, glucocorticoid treatment (0.5–1 mg/kg/day of predniso(lo)ne alone) in patients with acute severe AIH was effective and did not significantly increase the risk of infection[185,186,188,189]. Glucocorticoids treatment may be considered in patients with ALF due to AIH, being cautious of complications, such as infection [187]; however, in particular, patients with a Model for End-stage Liver disease (MELD) score >40 had lower overall survival with glucocorticoids treatment [190]. When treating acute severe AIH with glucocorticoids, it is important to promptly evaluate the clinical course and treatment response within 1 to 2 weeks and proceed with liver transplantation if there is little effect. Liver transplantation should be considered without delay if liver enzyme levels do not decrease, clinical symptoms worsen, or hepatic encephalopathy progresses [184,186,191]. Accompanying hepatic encephalopathy means ALF, and in this case, liver transplantation is more helpful in prognosis than glucocorticoids treatment [184,187].

Clinical parameters of treatment withdrawal

The goal of AIH treatment is to reduce mortality and increase survival by continuously controlling disease activity. However, it is practically difficult to determine the end of immunosuppressive therapy after the achievement of treatment goals. Therefore, alternative clinical parameters that are readily measurable and reflect the achievement of treatment goals are needed when considering the withdrawal of treatment. In clinical practice, complete normalization of serum transaminases and IgG levels for at least 2 years have been proposed as requirements before attempting treatment withdrawal [18,21,192]. Histologic examination prior to treatment withdrawal has been the preferred strategy, as histologic features are predictors of fibrosis progression and relapse. However, in adult patients with and without treatment withdrawal liver biopsy guidance, the rates of relapse were similar in both groups (30% vs. 21%) after treatment for at least 2 years following complete biochemical remission [193]. Of 28 patients in biochemical remission for at least 2 years before treatment withdrawal, 15 (54%) patients remained in biochemical remission after withdrawal, and five of 11 (46%) patients who showed histologic remission subsequently relapsed during a median follow-up of 28 months [192]. Since liver biopsy is not always available in clinical practice, it may not be mandatory before treatment withdrawal in all adult patients [20,21]. However, for patients with poor adherence to treatment or severe clinical symptoms, a liver biopsy should be considered prior to treatment withdrawal [18].

Non-invasive assessment of fibrosis by transient elastography might aid in the decision of treatment withdrawal. In a recent study, liver stiffness decreased by 7.5% per year in patients with biochemical remission, whereas those who were not in biochemical remission showed an increase in liver stiffness by 1.7% per year [135]. The average liver stiffness measurement of pre- and post-treatment was 8.2±6.7 kPa and 6.4±3.2 kPa in the remission group, and 8.1±5.8 kPa and 9.2±9.1 kPa in the non-remission group, respectively. However, a cutoff of liver stiffness for predicting remission or the association between liver stiffness and long-term outcome after treatment has not been determined yet. Therefore, its role in predicting relapse after withdrawal is unknown, and further studies are required.

Follow-up after treatment withdrawal and treatment of relapse

Relapses are very common after treatment withdrawal in AIH patients. Even patients with complete biochemical response for ≥2 years have shown relapse rates of 20–46% [192,194]. Most relapses occur within 6 to 12 months. About 50% of patients develop relapse within the first 3 months, and the frequency decreases after the first year to about 3% per year [195]. However, since late relapse also can occur, patients should be closely monitored in 3- to 6-month intervals for the first 1 year after treatment withdrawal and in 6-to 12-month intervals thereafter [20,21].

Relapse is usually asymptomatic, manifested by mild elevation of serum transaminases [196]. However, delayed or failed detection resulted in fibrosis progression in 10% of cases, and deterioration of hepatic function in 3% [197]. Therefore, early detection and prompt treatment of relapse is required. A liver biopsy is usually not mandatory to confirm the relapse, as the elevation of serum ALT is highly predictive. Various factors have been proposed as factors associated with relapse, such as slow response to treatment and short treatment duration [192,193], psychological stress or concomitant autoimmune diseases [198,199], AST or ALT levels above ULN or high IgG levels (>1.5 g/dL) at discontinuation of treatment [146,152,192,200], infiltration of plasma cells in the portal area at discontinuation [201], and prednisolone monotherapy [169], while the presence of clinical or histological cirrhosis at initial diagnosis or treatment withdrawal was not associated with relapse [194,197,199,202].

Treatment of the relapse aims to resume the initial treatment which led to remission [196,199,202-204]. Treatment should be initiated at induction doses, followed by glucocorticoid tapering. Once biochemical remission is re-established, the dose of AZA can be increased to 2 mg/kg daily, as glucocorticoid is reduced to the lowest dose needed to maintain remission or fully withdrawn. [204,205]. Patients intolerant of AZA can be treated with MMF or the lowest dose of glucocorticoid monotherapy needed to maintain biochemical remission [204,206].

Bone density assessments

Osteoporosis is associated with fractures which may lower the health-related quality of life. The initial fracture risk assessment or bone mineral density test should be performed before initiating glucocorticoid treatment, since the use of glucocorticoid can cause osteoporosis. The risk of osteoporotic fracture increases in patients receiving more than 7.5 mg of prednisolone daily or a cumulative dose of at least 5 g in the past year [207].

The Korean glucocorticoid-induced osteoporosis guideline recommends bone mineral density (BMD) testing within 6 months of the initiation of glucocorticoid treatment if patients aged <40 years have a history of osteoporotic fracture or other risk factors for osteoporosis (thyroid disease, history of smoking or alcohol use, etc.). It also recommends assessing the fracture risk using FRAX (Fracture Risk Assessment Tool, https://www.sheffield.ac.uk/FRAX/tool.aspx) and measuring BMD for patients aged ≥40 years within 6 months of the initiation of glucocorticoid treatment [208].

The risk of fracture should be reassessed every year if glucocorticoids are used continuously [208]. FRAX and BMD reassessment should be performed every 1 to 3 years for patients ≥40 years of age who are taking glucocorticoids continuously, but not treated with osteoporosis medications beyond calcium and vitamin D [208]. If patients <40 years of age are taking high dose of glucocorticoids (prednisolone ≥30 mg/day and cumulative dose >5 g/year) or have a history of osteoporotic fracture, Z-score <-3 at hip or spine BMD, or other osteoporosis risk factors, BMD test should be repeated every 2 to 3 years [208].

According to the Korean glucocorticoid-induced osteoporosis guideline, patients are recommended to take calcium (1,000–1,200 mg/day) and vitamin D (800 IU/day), and maintain adequate vitamin D concentrations (≥20 ng/mL) [208].

About 30% of patients with chronic liver disease are known to have osteoporosis regardless of the use of glucocorticoid [209]. The European guideline on nutrition in chronic liver disease recommends repeating BMD evaluation after 2 to 3 years even in patients with normal BMD. Supplementation of calcium (1,000–1,500 mg/day) and vitamin D (400–800 IU/day) is also recommended, although there is insufficient data confirming that these supplements can prevent bone loss in patients with liver disease [209].

Viral hepatitis assessments

Vaccination or infection status of viral hepatitis should be reviewed before initiating immunosuppressive therapy for AIH. If HBV infection status is unclear, screening for hepatitis B surface antigen (HBsAg) and anti-HBc IgG is necessary prior to immunosuppressive therapy. If either HBsAg or anti-HBc IgG is positive, hepatitis B virus (HBV) DNA test should be performed [210].

Vaccination against hepatitis A virus (HAV) and HBV should be given to patients with AIH, since hepatitis A or hepatitis B can increase morbidity and mortality in patients with preexisting chronic liver disease [211]. Accordingly, HAV and HBV vaccination is recommended for patients with AIH [20,21]. HAV vaccination should be given in a 2-dose series at 0 and 6–18 months to patients <40 years of age regardless of the test status for antibody to HAV or seronegative patients ≥40 years of age [212]. If HBsAg and anti-HBs are negative, HBV vaccination of 3-dose series should be administered on a schedule of 0, 1, and 6 months for patients who have not been vaccinated [213].

Out of 15 patients with autoimmune liver diseases (10 patients receiving immunosuppressive therapy), 100% developed anti-HAV after vaccination. For HBV, 16 (eight patients receiving immunosuppressive therapy) out of 21 patients (12 patients receiving immunosuppressive therapy) developed antibody against HBV after vaccination; the response rate (76%) was rather low, considering that the protective antibodies are generally detected in more than 95% after HBV vaccination [214,215]

Genotyping

AZA, one of the thiopurines, can cause adverse effects such as myelosuppression. Pretreatment testing for genotypes associated with drug metabolism may predict the incidence of myelosuppression; TPMT and NUDT15 polymorphisms are well-known for their association, respectively.

Other adverse events

Side effects of glucocorticoid include Cushingoid face, weight gain, osteoporosis, diabetes, cataracts, psychosis, and high blood pressure. Cushingoid face and buffalo hump occur in about 50% of patients; diabetes in 15–20%; and hypertension, psychosis, cataracts, and osteoporosis in about 5–10% [155-157,225]. Although adverse effects are observed less frequently after combined treatment with AZA than with glucocorticoid alone, at least 5% of patients experience adverse effects of glucocorticoids [163]. To detect these side effects, periodic physical measurements, blood pressure measurements, and blood tests are required. There is a meta-analysis in which the risk of gastrointestinal bleeding and perforation increases only in hospitalized patients using glucocorticoids [226], but the role of preventive drug administration has not been proven.

The side effects of AZA include nausea, heartburn, hepatotoxicity, hair loss, loss of appetite, fatigue, bruise, bone marrow suppression, increased infection vulnerability, and increased cancer risk. About 25% of patients receiving AZA treatment suffer side effects, and about 10% of patients need to stop the AZA treatment. These side effects are more common in patients with cirrhosis. In about 5% of patients, severe side effects, such as joint pain, fever, skin rash, and pancreatitis, occur within days or weeks of treatment, which results in discontinuation of the drug, and this reaction disappears within several days [20]. Since bone marrow suppression can occur, AZA treatment is not recommended in patients with patients with severe leukopenia (<2.5×10⁹/L), severe thrombocytopenia (<50×10⁹/L), or homozygote mutation in the NUDT15 or TPMT gene. The frequency and intensity of the side effects of AZA depend on the dosage and duration of treatment, the type of combination therapy, and the underlying conditions [227]. A rare case of AZA-induced pneumonia was reported in patients with ulcerative colitis, accompanied by symptoms such as fever, respiratory failure, pulmonary nodular opacity, and ground glass shadowing were observed; and these symptoms disappeared after discontinuation of the drug [228].

If re-use of AZA is inevitable, it can be referred that 64% (9/14) of patients could be re-administered AZA by restarting with dose escalation in a report of patients with inflammatory bowel disease who stopped AZA treatment due to flu-like symptoms, such as severe myalgia, headache, and fever [229].

Second-line agents

Mycophenolate mofetil (MMF): MMF is an inosine-5’-monophosphate that decreases both T-cell and B-cell proliferations, and it suppresses antibody formation and cell-mediated immunity [231]. MMF (1–2 g/day) is given with glucocorticoids for patients intolerant of AZA or who have incomplete or no response to first-line treatment. An Australian multicenter retrospective study demonstrated that 3-month biochemical response was 61.0% (57.1% in treatment intolerance and 61.9% in suboptimal response) in 105 patients who received MMF after first-line treatment (42 patients with drug intolerance and 63 patients with incomplete or no response to standard therapy) [232]. A meta-analysis showed that the biochemical response rate was 58.0–59.6% in patients who received MMF as a second-line treatment for AIH (73.5–82.0% in drug intolerance and 32.0-40.8% in incomplete or no response to standard therapy) [233,234].

Adverse events of MMF include nausea, vomiting, diarrhea, heartburn, headache, dizziness, skin rash, and infection. MMF is contraindicated in pregnant patients due to its teratogenic effect [235]. Adverse events are observed in 14.0–24.1% of patients receiving MMF. The most common adverse event is leukopenia, which can be controlled by dosage reduction, followed by gastrointestinal symptoms such as nausea, vomiting, and diarrhea. Rarely, severe leukopenia, sepsis, myalgia, pancreatitis, headache, hair loss, and paresthesia may also develop [233,234].

Cyclosporine: Cyclosporine, a calcineurin inhibitor, can be given in combination with glucocorticoids as a second-line treatment for AIH (2–3 mg/kg/day). The biochemical response of cyclosporine was 80% in five patients who showed incomplete or no response to glucocorticoid and AZA therapy [236]. Adverse advents, such as nephrotoxicity and changes in facial appearance (hypertrichosis), are commonly observed. However, evidence is lacking to support the use of cyclosporine as a second-line treatment.

Tacrolimus: Tacrolimus is a more potent calcineurin inhibitor that has a lesser change in facial appearance than cyclosporine. Tacrolimus is used in combination with glucocorticoids as a second-line treatment (0.5–6 mg/day). A European retrospective study showed that a biochemical response with tacrolimus as a second-line treatment was 52.9% in 17 AIH patients with intolerance to AZA treatment (n=1) and incomplete response to standard treatment (n=16) [237]. A meta-analysis revealed that the biochemical response rate was 68.9% in AIH patients who received tacrolimus as a second-line treatment (56.6% with intolerance and 59.1% with incomplete or no response to standard therapy) [234]. Adverse events of tacrolimus include neurologic symptoms, gastrointestinal symptoms, nephrotoxicity, diabetes mellitus, hypertension, and hair loss [238]. Nephrotoxicity can occur frequently in patients with high blood levels of tacrolimus. Therapeutic drug monitoring of tacrolimus is necessary since between-individual variability, drug-drug interactions, and drug-food interactions may affect the blood tacrolimus levels [239]. Adverse events of tacrolimus were reported in 25.5% [234].

Few studies compared MMF and tacrolimus as second-line treatments for AIH. A global retrospective study analyzed a total of 201 patients receiving MMF (n=121) or tacrolimus (n=80) as a second-line treatment for AIH, which consisted of 108 patients with intolerance and 93 patients with incomplete or no response to first-line treatment [240]. No significant differences were observed in the complete response rate between the two groups (MMF, 69.4% vs. tacrolimus, 72.5%; P=0.639). However, the complete response rate to tacrolimus (56.5%) was higher than that to MMF (34.0%) in patients who showed incomplete or no response to the standard therapy (P=0.029). Adverse events and liver-related mortality were not significantly different between the two groups [240]. However, it needs to be validated with prospective studies since the study might include several biases caused by the retrospective manner. The practice guideline for AIH released by the American Association for the Study of Liver Diseases (AASLD) in 2019 performed a meta-analysis on the comparison between MMF and tacrolimus as the second-line treatments for AIH. Although the meta-analysis revealed that both of the two drugs were effective in AIH as the second-line treatments, the AASLD guideline preferred MMF to tacrolimus considering the drug accessibility and safety profile [21]. However, further studies are needed to clarify due to a lack of high-quality evidence.

6-Mercaptopurine, 6-thioguanine, allopurinol: 6-MP can be administered in patients with intolerance to AZA for AIH. 6-TGN, an active metabolite, is derived from 6-MP, which is the first metabolite of AZA. Although 6-MP is not more effective than AZA, the use of 6-MP can improve medication adherence if there is intolerance to AZA (25–50 mg/day) [241]. In a study in which AZA was changed to 6-MP as a second-line treatment for a total of 22 patients after combination therapy of glucocorticoid and AZA, 75% of 20 patients with AZA intolerance showed biochemical remission, but two patients who had an inadequate response to AZA treatment did not respond [242]. Adverse events of 6-MP include leukopenia, headache, nausea, vomiting, diarrhea, oral ulcer, skin rash, and arthralgia.

6-Thioguanine (6-TG) is metabolized into 6-TGN, active metabolite of AZA. 6-TG can be given to patients with intolerance to AZA for AIH (20 mg/day). A biochemical response with 6-TG as a second-line treatment was 58% in 49 Dutch patients who had intolerance or insufficient response to AZA or 6-MP [243]. Adverse events of 6-TG include nausea, vomiting, poor oral intake, oral ulcer, and headache. 6-TG should be used with caution because of its hepatotoxicity. A meta-analysis showed that hepatic sinusoidal obstruction syndrome developed in 9–25% of 4,849 patients who received 6-TG for inflammatory bowel disease or acute lymphocytic leukemia. The risk of hepatic sinusoidal obstruction syndrome increases with the above daily dosage of 6-TG 25 mg [244].

Allopurinol (100 mg/day) can be added to AZA and glucocorticoid in AIH patients with intolerance to AZA or non-response. The immunosuppressive effect of AZA decreases if it is not metabolized to 6-TGN, active metabolite, but to 6-thiouric acid (6-TU) through xanthine oxidase [235]. Allopurinol can be used in AIH by inhibiting xanthine oxidase that induces AZA to 6-TGN. Low dosage of AZA with allopurinol may be one of treatment options as the second-line treatment. Although, studies on the use of allopurinol for AIH are insufficient, a study reported that the biochemical response rate of adding allopurinol in eight patients who failed with AZA was 88% [245].

Sirolimus, everolimus: Sirolimus and everolimus, mammalian target of rapamycin (mTOR) inhibitors, block IL-2-mediated signal transduction. Subsequently, this decreases the response of T-cell activation by cytokines and the formation of antibody, which prevents cell-cycle progression and proliferation [246]. Further studies are warranted to clarify the use of sirolimus or everolimus as the second-line treatments for AIH, as previous studies have been conducted only for a small number of patients [247,248].

Infliximab, rituximab: Infliximab, a tumor necrosis factor (TNF)-alpha inhibitor, has been tried as a second-line treatment. Of 11 patients receiving infliximab for non-response to treatments, including AZA for AIH, AST/ALT and serum IgG levels were normalized in eight patients and six patients, respectively. However, infection-related complications were observed in seven patients [249]. TNF-alpha antagonists, including infliximab, should be used with caution as they can increase the risk of DILI, including AIH-like DILI [250,251].

Rituximab is a monoclonal antibody against CD20, a surface antigen of B lymphocyte. Rituximab is firstly introduced to treat B-cell lymphoma and has been tried to treat a variety of autoimmune diseases. A phase 1 clinical trial on the use of rituximab was conducted for six AIH patients with treatment failure to glucocorticoid and AZA. After the administration of rituximab 1,000 mg with two sessions at 2-week intervals, all of the patients had decreased levels of AST, ALT, and serum IgG, and then a reduction in the glucocorticoid dosage. Serious adverse events were not observed [252]. However, further studies on rituximab for AIH are warranted.

Treatment indications

AIH should be suspected and tested if liver disease is suspected in children without other infectious or metabolic causes. If AIH is diagnosed, treatment should be initiated immediately, as it is important to prevent the progression of the disease [32]. In most cases of AIH, except for those with liver failure and hepatic encephalopathy, the remission rate of immunosuppressant treatment is as high as 90%, regardless of the degree of liver damage [37,41,145,253]. Cirrhosis is observed in 40–88% of pediatric patients with AIH at the time of diagnosis [41,254,255]; however, most pediatric patients receive long-term treatment since the mortality rate is low at this age.

Treatment aims

The goals of AIH treatment are to reduce or eliminate liver inflammation, induce complete biochemical response, improve symptoms, and increase life expectancy in the long term [32]. To achieve these goals, most of the patients need continuous maintenance treatment, and a few patients can maintain remission after treatment withdrawal [20]. The ideal complete biochemical response after treatment is the normalization of serum AST, ALT, and IgG [21,144,145]. However, in children, the criterion that negative conversion or low titer of autoantibodies (ANA & SMA <1:20, anti-LKM1 & anti-LC1 <1:10) and histological loss of inflammation should also be considered [32,256]. This is because histological response lags behind biochemical response, and clinical or biochemical remission does not necessarily reflect the histological loss of inflammation [256].

First-line treatments

The traditional first-line treatment for pediatric AIH is the combination of prednisolone and AZA (Fig. 7) [20,21,32,154-157,163,257-260]. Prednisolone should be started at 1–2 mg/kg/day (maximum 60 mg/day) and then tapered over 4 to 6 weeks, and maintained at a dose of 2.5–5 mg/day as aminotransferase levels decrease. Although the timing of AZA initiation is controversial, it is recommended to add prednisolone around 2 weeks after treatment to accurately evaluate the treatment response to prednisolone and exclude AZA-induced hepatotoxicity [21]. AZA is started at the dose of 0.5 mg/kg/day, maintained at 1–2 mg/kg/day, and can be increased to 2–2.5 mg/kg/day until a complete biochemical response is achieved in the absence of toxicity [261]. AZA should be withheld until liver function improves, and prednisolone alone should be given in patients with decompensated cirrhosis or liver failure since AZA may cause hepatotoxicity [21].

Figure 7.

Treatment algorithm for pediatric autoimmune hepatitis. AIH, autoimmune hepatitis; MMF, mycophenolate mofetil; TNF, tumor necrosis factor; IgG, immunoglobulin G. ^*Do not use in decompensated cirrhosis or liver failure. ^†Second or third-line treatment should be initiated and monitored only in specialized centers.

Aminotransferase levels usually decrease by 80% or more within 2 months, but it may take several months to fully normalize. In the beginning, liver function tests are performed every 1–2 weeks, and prednisolone can be tapered after achieving a biochemical response. For 4 to 8 weeks after initiation of treatment, the dosage should be adjusted while monitoring biochemical response and drug adverse events every 2 to 4 weeks (Fig. 7).

Children often experience side effects, such as growth failure, weight gain, hypertension, hyperlipidemia, and osteoporosis, as prednisolone is administered for a relatively long period of time [41]. Therefore, prednisolone should be continuously tapered to the lowest dose that can maintain a complete biochemical response or stopped [262,263]. In some pediatric patients, complete biochemical response can be maintained with AZA monotherapy without glucocorticoids [172,261,264,265]. In a study that reduced the prednisolone dose or administered it every other day, the decrease in growth rate improved, but the final height of the patients was shorter than predicted [262]. However, recent studies suggested that appropriate treatment of glucocorticoid in pediatric AIH did not affect the final height. A long-term follow-up study of 28 Egyptian children with AIH reported that the final height was significantly affected by the AIH severity at the onset of the disease, and not by the continuation or the duration of corticosteroid treatment [266]. In another retrospective study of 75 pediatric patients, when low-dose prednisolone (2.5 mg under 12 years old, 5 mg over 12 years old) was prescribed and followed up for about 11 years, the patients’ Z-scores (height, weight, and body mass index) were consistently maintained within the normal range [267]. A subgroup analysis of a randomized controlled trial reported that the remission rate was similar between pediatric patients taking budesonide or prednisone as an induction therapy; however, the weight gain was lower in the budesonide group [268]. Therefore, budesonide can be considered, but oral budesonide is not currently available in South Korea.

The measurement of TPMT activity before starting AZA treatment is recommended in Western guidelines since it may prevent some adverse events [21,32]. However, mutations for NUDT15 rather than TPMT should be considered in Asia [224]. Since most studies were conducted in patients with inflammatory bowel disease, further research is needed for pediatric patients with autoimmune diseases [269].

Second-line treatments

Second-line treatments of AIH can be attempted when patients are intolerant of or do not respond to standard therapy. Drugs used as second-line treatments in children are listed below.

Budesonide: In a double-blind, randomized controlled study of pediatric patients with AIH, budesonide was effective in inducing remission and had fewer side effects than prednisolone [268]. Since more than 90% of budesonide is first transferred to the liver and metabolized, it has few systemic side effects and is also called the local treatment of the liver [270]. However, it cannot be used in patients with LC or poor liver function, and it is not marketed as an oral medication in South Korea.

Mycophenolate mofetil (MMF): In a retrospective study, MMF was used in combination with corticosteroids at doses of 20–40 mg/kg twice daily in 26 children with AIH who did not respond to the first-line treatment. Among them, 18 children responded to MMF, and AST normalized in 14 patients. Six of the eight patients who did not respond to MMF had autoimmune sclerosing cholangitis [271]. The most common adverse event was leukopenia, but there were no serious side effects. In a single-center prospective observational study including two 16- and 17-year-old children, all two children showed normalization of ALT when using MMF [272]. However, the second-line treatment of MMF still remains controversial due to some studies demonstrating the poor effect of MMF, especially in children who do not respond to AZA [37,273-275].

According to a meta-analysis of second-line treatments in pediatric AIH patients who did not respond to the standard therapy, the response rate was the highest with cyclosporine; however, the adverse event rate was also the highest. MMF showed the second-highest response rate and a low adverse event rate [276]. Calcineurin inhibitors can be considered as an alternative option if there are persistent adverse events to MMF (headache, diarrhea, nausea, hair loss, leukopenia, etc.) [32,276].

Cyclosporine: In a Canadian study, treatment-naïve pediatric AIH patients were treated with cyclosporine (4 mg/kg/day in 3 divided dose) alone for 6 months, and then maintained with prednisolone and AZA. Among them, 72% of patients showed normalization of ALT level and improvement in growth rate within 6 months of treatment, with few side effects [277,278]. According to these results, cyclosporine is used as a first-line treatment for AIH in children in some regions, but the level of evidence is still weak [32]. A retrospective study was performed to evaluate the effect of cyclosporine for 13 pediatric patients with type 2 AIH. The patients were treated with cyclosporine due to intolerance (n=8) and non-response (n=5) to the first-line treatment. When cyclosporine was used at a blood concentration of 200–250 ng/mL at the beginning of treatment and maintained at a blood concentration of 100–150 ng/mL for 1 year after remission, the ALT levels were normalized within 6 months in both groups [279]. Additional studies on successful maintenance therapy using cyclosporine in children and adolescent patients were reported [280,281]. In a randomized trial comparing the combination therapy of prednisolone and AZA to cyclosporine in 26 pediatric patients, the treatment effect and safety of the combination therapy and cyclosporine were similar [282].

Tacrolimus: Tacrolimus is a calcineurin inhibitor that is more effective and has fewer side effects than cyclosporine, a drug with the same mechanism. Although the use of tacrolimus has not been studied much in pediatric AIH patients, a non-randomized prospective study revealed that tacrolimus showed an effect of reducing the dose of prednisolone and AZA when used at a target blood concentration of 2.5–5 ng/mL in 17 pediatric patients. However, tacrolimus monotherapy did not induce a complete biochemical response, and adverse events such as headache, abdominal pain, chronic inflammatory bowel disease, and deterioration of liver function occurred [283]. A multinational retrospective study compared the effects of MMF and tacrolimus in 38 pediatric patients who were intolerant to corticosteroids or AZA and non-responsive to the first-line treatment [284]. Complete biochemical response was seen in 88.9% of MMF and 87.5% of tacrolimus in the intolerant group. In the non-response group, a complete response rate was 22.5% in MMF and 50% in tacrolimus. There was no statistical significance between the two drugs [284].

Rituximab, infliximab, sirolimus: Rituximab is an anti-B lymphocyte monoclonal antibody. According to a case report on children with AIH who do not respond to first-line treatment, MMF, and cyclosporine, there was a decrease in liver enzyme levels and IgG was maintained for 12 to 26 months after rituximab treatment (every 4 weeks to 6 months intravenously). No serious side effects were observed [285].

Infliximab may be attempted as a rescue treatment, as some case reports have shown its effect on refractory pediatric AIH [286,287]. However, it has been reported that immune-related hepatitis occurred after infliximab treatment in some pediatric patients with inflammatory bowel disease, and the role of TNF-a in the development of AIH is not fully known. Therefore, infliximab should be used with caution for refractory AIH [288].

Sirolimus is a drug that selectively expands regulatory T cells. In a case report of four pediatric non-responsive patients with AIH, sirolimus was treated and two patients responded to sirolimus. A small series in adult patients with refractory AIH also showed similar results; however, the evidence for sirolimus is insufficient to date [247,289].

Treatment withdrawal

Complete biochemical response, including the normalization of AST, ALT, IgG levels, and negativity or low titer of autoantibodies (ANA & SMA <1:20, anti-LKM1 & anti-LC1 <1:10) should be maintained for at least 2–3 years before treatment withdrawal is tried in children with AIH [32]. Although liver biopsy before the treatment withdrawal is not essential in adults, it is still recommended in children [21,32,41,290].

Most of the studies on treatment withdrawal were conducted retrospectively. In a recent prospective study including adults and children, the treatment was terminated when patients showed normal ALT and IgG levels for at least 2 years in addition to histological remission. The remission maintenance after treatment withdrawal was seen in 67% during the 62-month observation period, which was higher than what was observed in previous studies. However, the number of patients in this study was only 12, and all of them were type 1 AIH patients known to have a relatively low recurrence [291]. Other retrospective studies, including children, reported that histologic findings did not predict recurrence [292].

In a prospective study observing children with AIH for 16 years, treatment withdrawal was attempted in seven children with AIH who showed normal liver enzyme levels for at least 1 year and had no inflammation in follow-up liver biopsy. Among them, three patients were able to stop treatment completely, and three patients with type 1 AIH could discontinue prednisolone and reduce the AZA dose. However, one patient with type 2 AIH recurred 8 months after the treatment withdrawal [293]. In a pediatric study conducted in 1984, treatment withdrawal was attempted after liver biopsy in nine patients. Among them, three patients with type 1 AIH and five patients with type 2 AIH relapsed. In eight patients who relapsed, three patients showed remaining histological inflammation, but five patients confirmed histological loss of inflammation before the withdrawal [262]. Since recurrence of pediatric AIH is relatively common, it is important to prevent recurrence by carefully attempting to withdraw treatment after confirming the complete disappearance of inflammation in liver histology as well as a complete biochemical response.

Pregnancy

Pregnancy induces changes in the immune system, which can affect the course of AIH [294]. Pregnancy requires maternal immune tolerance to paternal alloantigens expressed in fetal tissues [295-297]. In general, the activity of autoimmune diseases decreases during pregnancy, and after childbirth, the activity of autoimmune diseases increases, thereby increasing the risk of exacerbation of autoimmune diseases [297,298]. According to studies reporting the clinical course during pregnancy in patients with AIH, serum ALT levels usually decreases during pregnancy, and spontaneous remission during pregnancy is reported in some cases. The risk of acute exacerbation is high during the first 3 months after childbirth [299-302]. However, the course of AIH during pregnancy is very diverse. Acute flares and emergent liver transplantation during pregnancy have been reported [299,300,303]. In particular, the risk of acute exacerbation was high in cases of poor drug compliance, such as discontinuing the drug or reducing the dose against medical advice before or during pregnancy [304]. Therefore, if a patient with AIH is considering pregnancy or becomes pregnant, it is necessary to carefully check for drug compliance, be aware of the possibility of acute exacerbation, and closely conduct follow-ups before, during, and after pregnancy. Not receiving immunosuppressive treatment during pregnancy and having a short remission period (less than 1 year) before pregnancy have been reported as risk factors for acute exacerbations during pregnancy and after childbirth [299,304,305].

AIH and treatment for AIH can affect pregnancy outcomes [294]. There was no difference in the live birth rate between the general population and mothers with AIH [7,306]. However, obstetric complications in patients with AIH were higher than that in the general population [307,308]. According to the analysis of 18,595,345 cases of maternal discharge data from 2012 to 2016 in the United States, the risk of postpartum hemorrhage and maternal death was the same for mothers with AIH compared to the general population, but the risks of gestational diabetes (OR, 2.2; 95% CI, 1.5–3.9), pre-eclampsia and eclampsia (OR, 2.4; 95% CI, 1.6–3.6), and preterm birth (OR, 2.0; 95% CI, 1.2–3.5) were higher for mothers with AIH compared to the general population [307]. In a meta-analysis analyzing the clinical outcome of mothers diagnosed with AIH, the risk of preterm birth was higher than that of the general population (RR, 2.45; 95% CI, 1.66–3.62) [301]. In particular, the risk of preterm birth was high in mothers with portal hypertension or in mothers without biochemical remission before pregnancy [301], and patients with cirrhosis had a higher risk of liver failure or need for a liver transplant [299].

Various drugs used in AIH treatment can affect fetal outcomes [309]. Recent studies have reported no significant association between glucocorticoids and fetal malformations, but glucocorticoids have been reported to increase the risk of cleft lip in the fetus [310,311]. AZA has been reported to be associated with fetal malformations in animal experiments; however, in humans, the association with fetal malformations was not found [312,313]. On the other hand, MMF has been associated with teratogenicity in humans [312]. Therefore, MMF use is contraindicated in pregnancy, and women are recommended to conceive after discontinuing MMF for at least 6 weeks [312]. Glucocorticoids and AZA can be used during pregnancy and lactation [309].

A shorter duration of remission before pregnancy increases the risk of acute exacerbation of AIH during pregnancy [299,305]. Therefore, if possible, it is recommended that pregnancy be considered after the disease has been well-controlled for at least 1 year. For patients planning a pregnancy, the drug contraindicated during pregnancy should be discontinued or changed. If pregnancy is confirmed for a patient receiving glucocorticoids and/or AZA therapy, the drug dose should be minimized to the dose that can maintain remission during pregnancy. During pregnancy, close monitoring for disease activity is required, and special attention is needed especially during the early period after childbirth, as the immunosuppressant requirements to maintain remission can change. During pregnancy, a reduction in immunosuppressant dose can be considered. If immunosuppressants have been reduced during pregnancy, the dose can be preemptively increased after childbirth to the pre-pregnancy dose. The first 3 months after childbirth is a period of high risk for acute exacerbations [299,302], and careful follow-up at short intervals is necessary. The use of glucocorticoids and AZA is not contraindicative to breastfeeding.

Old age

AIH can occur at any age. However, AIH diagnosed in the elderly is characterized by more asymptomatic cases, more often delayed diagnosis, and higher rates of comorbidity, cirrhosis, and extrahepatic diseases, such as thyroid or rheumatic disease [60,314-316]. Elderly patients respond well to the immunosuppressive therapy, with less relapse after treatment withdrawal [317,318]. However, the cut-off age to define the elderly population varies from 60 to 70 years in each study, and some studies have reported no difference in clinical features and treatment responses according to age [315,319].

In elderly patients, drug use to control other comorbidities is common. In these cases, it is sometimes very difficult or impossible to differentiate DILI from AIH [320,321]. DILI and AIH may be differentiated through liver biopsy [322], by monitoring the clinical course after discontinuation of suspected drug, or by monitoring the clinical course after reducing or discontinuing glucocorticoids if glucocorticoids have been used [120].

Elderly patients have a higher rate of comorbidities, such as osteopenia, hypertension, and diabetes, and a higher rate of liver cirrhosis, which can prevent them from receiving immunosuppressive therapy [319]. Immunosuppressive therapy, especially glucocorticoid treatment, can affect the clinical course of comorbidities, such as osteopenia, hypertension, and diabetes. Therefore, when AIH is diagnosed in an elderly patient, comorbidities should be evaluated, and in particular, osteopenia and osteoporosis should be checked in advance [323]. Careful follow-up without immunosuppressive treatment can be considered in elderly patients if AIH is asymptomatic, comorbidity is severe, and there is a high possibility of aggravation of comorbidity after immunosuppressive treatment. However, there is no reason to withhold immunosuppressive treatment just because of age.

AIH overlap syndromes

Overlap syndrome refers to cases that meet the diagnostic criteria for AIH and diagnostic criteria of other autoimmune liver diseases, mainly PBC or PSC. AIH-PBC overlap syndrome has a higher risk of cirrhosis complications than AIH alone or PBC alone [324-326]. In a Korean study, AIH-PBC overlap syndrome showed poorer clinical outcomes compared to AIH alone or PBC alone [50]. AIH-PSC overlap syndrome shows more severe disease and worse prognosis compared to AIH or AIH-PBC overlap syndrome [327].

Clinical presentations and clinical courses are very heterogeneous in patients with AIH [51]; therefore, the treatment of overlap syndrome should empirically focus on predominating features [49]. In a retrospective analysis of a randomized controlled trial evaluating ursodeoxycholic acid (UDCA) response in patients with PBC, 16 patients showed AIH features as well (ALT >5xULN; IgG >2xULN or positive SMA; and moderate to severe lobular inflammation on pretreatment liver biopsy) [328]. In these patients, response to UDCA therapy was similar to those of patients with PBC without features of AIH, suggesting that UDCA therapy without glucocorticoids can be considered for AIH-PBC overlap syndrome with predominant features of PBC [328]. However, the analysis was limited by small size of AIH-PBC patients receiving UDCA therapy. For patients with AIH-PBC overlap syndrome, the combination use of UDCA and glucocorticoid showed biochemical improvement for patients not responding to either UDCA alone or glucocorticoid alone therapy [106]. In a meta-analysis comparing UDCA alone and UDCA with corticosteroids and/or AZA, transplant-free survival was better in the combination treatment group [329]. Hence, UDCA treatment first and sequential addition of immunosuppressive drug can be considered for AIH-PBC overlap syndrome patients if the patient shows predominant features of PBC. Otherwise, combination treatment with UDCA with the immunosuppressive drug is preferred for AIH-PBC overlap syndrome. A previous study reported that immunosuppressive treatment can be helpful in AIH-PSC overlap syndrome, but it was limited by the observational nature of the study [330].

AIH with viral hepatitis

AIH patients with chronic viral hepatitis B (CHB) have a risk of reactivation of CHB during immunosuppressive therapy, including glucocorticoid. Prophylactic antiviral treatment is recommended if HBsAg or HBV DNA is positive [210]. The high-risk group in which prednisolone is taken ≥20 mg/day for ≥4 weeks and the moderate-risk group in which prednisolone is taken 10-20 mg/day for ≥4 weeks are the indications of prophylactic antiviral treatment. [210,331]. In low-risk cases, in which prednisolone <10 mg/day or antimetabolite such as AZA is taken, it is possible to monitor regularly without antiviral treatment [331]. If the patients of isolated anti-HBc positivity with negative HBsAg and HBV DNA take glucocorticoid or AZA, regular follow-up (HBsAg and HBV DNA every 1–3 months) could be considered. [210].

The aggravation of hepatitis during immunosuppressive therapy is less frequent in chronic viral hepatitis C (CHC) compared with CHB; but rarely, acute exacerbation of CHC may occur during immunosuppressive therapy. Therefore, it is reasonable to treat HCV with direct-acting antivirals before or concomitantly on immunosuppressive treatment. Drug-drug interactions of direct-acting antivirals with immunosuppressant agents should be considered [18].

AIH with non-alcoholic fatty liver disease

Non-alcoholic fatty liver disease (NAFLD) is referred to as a hepatic phenotype of metabolic syndrome, and it often accompanies the factors of metabolic dysfunction such as hypertension, diabetes mellitus, dyslipidemia, and visceral obesity. Since the prevalence of NAFLD is as high as 25% in South Korea, a considerable proportion of AIH patients might have NAFLD in common [332]. In a Japanese retrospective study, 17% of 1,151 patients with AIH had NAFLD. Patients with both AIH and NAFLD showed a lower ratio of female to male, older age, lower elevated ALT level, and lower frequency of glucocorticoid treatment compared to the patients with AIH only [333]. A Western study of 73 patients with AIH reported that 16% and 24% had non-alcoholic steatohepatitis and simple steatosis, respectively [334]. In an interim analysis from an ongoing large-scale multicenter retrospective study in the International Autoimmune Hepatitis Group, 21.6% of 583 AIH patients had NAFLD [335]. A study from the United States presented that the patients with both AIH and NAFLD had 2.5 times liver-related morbidity and 7.6 times higher liver-related mortality compared to the patients with AIH only [334]. Meanwhile, although the positivity of autoimmune antibodies was as high as 23–33% among patients with NAFLD, the proportion of AIH patients confirmed by liver biopsy was less than 2% [336,337]. Therefore, liver biopsy is required to ascertain AIH when autoimmune antibodies are positive in patients with NAFLD.

Treatment for AIH with NAFLD consists of a general approach to NAFLD and specific management for AIH considering NAFLD. A general approach to NAFLD includes weight loss through lifestyle modifications, such as exercise and reduction of intake calorie, and proper medications [338]. Low-dose glucocorticoid and rapid tapering can be specific management for AIH considering NAFLD, but data about this treatment strategy are limited.

Budesonide instead of prednisolone could be used in case of uncontrolled diabetes mellitus without cirrhosis to ameliorate the systemic adverse effects of glucocorticoid, but it is not available in South Korea [176]. Besides that, it can be tried to add AZA or MMF early up to a maximum tolerable dose, and then taper glucocorticoid within 6 months [339].

Recurrent or de novo AIH following liver transplantation

According to data analyzing 4,085 AIH patients registered during 2009–2013 in South Korea, 1.1% of AIH patients received liver transplantation [6,19]. Patients who underwent liver transplantation for AIH had good post-transplantation outcomes, and the 5-year survival and graft survival rates were reported to be 72% and 65%, respectively [340]. After transplantation for AIH, the recurrence rate of AIH was reported to be 10–50% [341], and the recurrence rate of AIH reported by a single institution in South Korea was 14.3% at 5 years [342]. If AIH recurs or de-novo AIH develops after liver transplantation, the clinical course after liver transplantation becomes worse [343]. Studies have shown that the maintenance of glucocorticoids can reduce the risk of AIH recurrence for patients with AIH who received liver transplantation [344]. However, other studies have shown that glucocorticoids can be successfully discontinued in AIH patients after liver transplantation [345]. Strategies for the immunosuppressive drugs in patients undergoing liver transplantation for AIH require further research. In cases of liver transplantation due to AIH, the rates of acute rejection, glucocorticoid resistance rejection, and chronic rejection are higher than those who received liver transplantation for other reasons [346,347] However, it is often difficult to distinguish between rejection and recurrence of AIH, since the criteria for recurrence of AIH in patients after liver transplantation have not yet been established; therefore, care must be taken in interpreting the data [348]. In patients who received liver transplantation for other reasons, AIH can be newly diagnosed during the clinical course, defined as de novo AIH [340,343]. In case of recurrence or new occurrence of AIH, the use of glucocorticoids in addition to calcineurin inhibitors is considered the basis of treatment [341], but optimal immunosuppressant use strategies have not been well-established and require further research.

AIH with hepatocellular carcinoma

In AIH patients diagnosed with HCC, the minimum dose of immunosuppressant that can maintain remission is required, but additional research is needed on the optimal immunosuppressant use strategy. In a study of liver transplant patients, it was reported that calcineurin inhibitors increased the risk of HCC after transplantation, and that the risk of HCC was reduced when mTOR inhibitors were used [349]. Another study demonstrated that MMF reduced the risk of recurrence of HCC after liver transplantation [350]. These results suggest that the risk of HCC development or recurrence may vary depending on the type of immunosuppressive agent. Based on such findings, some studies have suggested that MMF is preferred over AZA as an immunosuppressive treatment for AIH patients with HCC [351]. However, no study has evaluated whether MMF or mTOR-based immunosuppressant treatment can improve the prognosis of AIH patients with HCC compared to treatment with glucocorticoids and AZA.

Immune checkpoint inhibitor-based treatment is recommended for patients with advanced HCC as a first-line option [352]. Immune checkpoint Inhibitors can cause autoimmune-related adverse events [353]. In patients with autoim-

mune diseases, such as rheumatic disease or inflammatory bowel disease, the risk of autoimmune-related adverse events after using immune checkpoint inhibitors is higher, but adverse events are mostly controlled with glucocorticoids and immunosuppressants and show similar treatment efficacy compared to patients without autoimmune diseases [354,355]. Hence, in cases of mild or well-controlled autoimmune disease, and in cases where flares may not be life-threatening, immune checkpoint inhibitor-based treatment might be considered for patients with pre-existing autoimmune disease [356]. However, there is a lack of data to examine the safety and efficacy of immune checkpoint inhibitors in AIH patients with HCC, as those have been excluded in clinical trials evaluating immune checkpoint inhibitor-based treatment [356]. If autoimmune-related adverse events or AIH flares occur after checkpoint inhibitor-based treatment, it may lead to life-threatening complications, such as liver failure. Therefore, immune checkpoint inhibitor-based therapy should be carefully considered after evaluating the potential risks and benefits.