A1. Background

Autoimmune hepatitis (AIH) is a chronic inflammatory liver disease which, if untreated, often leads to cirrhosis, liver failure and death. The historical development of concepts in relation to AIH has recently been reviewed.1 Major advances were made in its management based on controlled trials performed in England and the USA in the 1970s and 1980s. Unfortunately, in recent decades there has been a dearth of controlled clinical trials. Indeed, a recent systemic review2 identified only 11 randomised controlled trials in AIH and, thus, many questions regarding the optimal management of this disease remain unanswered. Many promising newer immunosuppressive therapies await formal comparison with standard therapies and many important details in relation to the application of standard therapies remain unclear. In these guidelines we will try to describe the optimal management strategies in adults based on available published evidence including the American Association for the Study of Liver Diseases (AASLD) practice guidelines for diagnosis and treatments of AIH published in 20023 and recently updated.4

A2. Grading of evidence and strength of recommendations

These are based on the GRADE system.5

Categories of evidence:

• I At least one high-quality randomised controlled trial.

• II-1 Non-randomised trials.

• II-2 Cohort or case–control analytical studies.

• II-3 Case series, uncontrolled observations.

• III Opinions of respected authorities.

Grading of evidence:

• A: High quality. Further research unlikely to change confidence in estimate of effect.

• B: Moderate quality. Further research likely to influence confidence in estimate of effect and may change the estimate.

• C: Low quality. Further research is very likely to influence confidence in estimate of effect and is likely to change the estimate.

• D: Very low quality. Any estimate of the effect is uncertain.

Strength of recommendation:

1. Strong, based on grade of evidence, consensus of opinion and estimated benefit to patients.

2. Weak, based on poor quality evidence and/or divergence of opinion.

A3. Epidemiology and pathogenesis

The reported prevalence of AIH ranges from 10 to 17 per 100 000 in Europe and appears to be similar to that of primary biliary cirrhosis.6–10 No published prevalence data are available from the UK. AIH accounted for two of 121 patients presenting to a UK hospital with jaundice.11

AIH has been described in many ethnic groups and seems to be a worldwide disease.12–17 Women are affected 3–4 times more frequently than men. Although initially thought to be particularly prevalent in young women, the disease appears to affect all age groups and, in the UK, may actually be more common in older than in younger patients.18 19

Most cases of AIH have no identifiable precipitant. There have been occasional cases presenting shortly after documented infection with hepatitis A,20–22 hepatitis E,23 cytomegalovirus24 25 and Epstein–Barr virus.26 Sometimes the disease may be precipitated by a drug. Several cases of AIH have been associated with minocycline,27–31 interferon α,32–35 nitrofurantoin36–38 and, recently, with infliximab.39 There are anecdotal reports of associations with many other drugs including ezetimibe,40 interferon β,41 42 ornidazole,43 diclofenac,44 indomethecin,45 terbinafine,46 methyldopa,47 ranitidine,48 atorvastatin,49 fluvastatin,50 fibrates,51 adalimubab52 and after hepatitis A vaccination.53 AIH has also been reported after herbal medicines.54 55 However, many of these associations may be coincidental.

In a recent study of 261 patients with AIH,56 24 (9%) were associated with drug ingestion—either nitrofurantoin or minocycline in most cases. Atypically for drug-induced liver injury, many patients with drug-related AIH had been taking the drug from many months or years. Other features included absence of cirrhosis and of disease recurrence following withdrawal of immunosuppressive treatment.

The immunopathogenesis of AIH has also been reviewed in detail by Vergani et al.57–59 While not fully understood, it is possible to consider the pathogenesis of AIH as an interaction between a trigger such as a drug or virus and the environment in a genetically susceptible individual.60

B1. Modes of presentation

The possibility of AIH is raised by persistently abnormal serum alanine aminotransferase (ALT) and/or aspartate aminotransferase (AST) values, usually accompanied by hyperglobulinaemia. About 25% of patients with AIH are asymptomatic at diagnosis,10 19 61–64 including some with cirrhosis.65 Patients commonly present with fatigue and general ill health, anorexia and weight loss, sometimes dating back years. Nausea is often a prominent symptom and amenorrhoea is common. Joint pains, sometimes flitting, are reported in 30–60% of patients, although joint swelling is uncommon. Rarer features include a maculopapular skin rash and unexplained fever.61

The traditional view of AIH is that of a chronic disease, diagnosis of which in the past has required serum transaminase elevations over a 3–6-month period. However, in about 40% of cases AIH presents as ‘acute hepatitis’ with jaundice, often preceded by anorexia, nausea and influenza-like symptoms.19 61 66 Serum AST levels may be several thousands. Such patients, if treated promptly, have a good outlook.67 68

Up to 50% of patients, even with an insidious onset of disease, may be clinically jaundiced or report previous episodes of icterus.61 About 30% of patients have cirrhosis at presentation10 19 61 62 so some patients (especially the elderly19) may present with ascites, suggesting liver decompensation and/or a variceal bleed.

AIH sometimes presents with acute liver failure.69–73 Some patients classified as having cryptogenic or seronegative fulminant hepatitis are likely to be patients with an acute presentation of AIH. In one series of acute liver failure, 30% had serum autoantibodies.74 In some of these the aetiology of the acute liver failure was clearly not autoimmune and the autoantibodies may have been epiphenomena. However, five of the 15 patients with otherwise unexplained acute liver failure74 met the International Autoimmune Hepatitis Group (IAIHG) criteria for probable AIH (see section B2c).

When AIH presents as acute hepatitis the liver histology may be atypical: lobular hepatitis and centrilobular necrosis are common and cirrhosis less common.66 67 75 Furthermore, serum autoantibodies are sometimes absent initially but develop later. Other causes of acute hepatitis (viral, drug-induced, Wilson's disease) need to be carefully excluded. In these sometimes very ill patients a definitive diagnosis of AIH (see section C2) and institution of immunosuppressive treatment should not await demonstration of ‘chronicity’ by monitoring liver tests over weeks/months. Patients with liver failure should be referred to a liver transplant centre.

In 30–50% of patients AIH is associated with other ‘autoimmune’ diseases (table 1) which may point to the diagnosis. Its occasional co-presentation with coeliac disease10 82–85 is important to recognise because, otherwise, malabsorption of immunosuppressive medication may delay effective treatment.

The presentation of AIH may vary between different races. In some reports African-American patients were younger and had a higher prevalence of cirrhosis and liver failure at presentation than those of Northern European origin.17 Non-Caucasian patients had more cholestatic biochemical and histological features and a higher rate of non-response,13 which was not explained by differences in access to medical care. South American patients also tend to have more severe disease than their North American counterparts,99 whereas Japanese patients tend to have late onset disease that responds to less potent immunosuppression.100 101

B2. Diagnosis

C1. Who should treat?

Given a prevalence of 1/10 000, it is likely that there will be about 25 patients with AIH in an area served by an average UK district general hospital. Management of all patients in specialist liver units is not necessary. Ideally, patients should be under the supervision of a hepatologist or a gastroenterologist with an interest in liver disease. They should be monitored in a designated liver clinic, ideally with the help of a specialist gastroenterology or liver nurse. At the very least, consideration should be given to having all patients under the care of one or two designated consultants. Arrangements should be in place for regular monitoring of immunosuppressive therapy in either primary or secondary care (see section C5). Monitoring should be lifelong. There should be a forum for regular discussion of problematic patients. Finally, there should be easy access to (a) an immunology laboratory that can assay (and quantify results in titres of) all relevant serum antibodies; (b) a specialist liver histopathologist; (c) a hepatologist with expertise in the management of AIH; and (d) a liver transplant centre. Finally, there should be regular audit of outcomes.

C2. Should everyone be treated?

Patients with AIH and moderate or severe inflammation (defined as one or more of serum AST >5 times normal, serum globulins >2 times normal, liver biopsy showing confluent necrosis) should be offered immunosuppressive treatment because of the clear survival benefits in these patients, demonstrated in the trials discussed below. In patients not meeting these criteria, treatment should also be considered if (a) the patient has symptoms (at least on a trial basis to see if they improve); (b) in patients with AIH and established cirrhosis on liver biopsy, since these are adverse prognostic features 61 64 149 159; and (c) in younger patients, in the hope of preventing cirrhosis development over several decades.

The benefits of immunosuppressive treatment in asymptomatic older patients with mild interface hepatitis (Ishak necroinflammatory score 4–6137) on biopsy are not established and a decision not to treat might be justified, especially if there are relative contraindications to the use of steroids. Ten-year survival in such untreated patients with mild disease was 90% in one study160 and 67% in another study (of only eight patients).161 In another uncontrolled study, untreated asymptomatic patients had similar survival to those receiving immunosuppression. However, 25% of these patients developed symptoms.64 Therefore, if untreated, patients with mild AIH should be monitored and, if symptoms develop or if liver tests remain abnormal, repeat liver biopsy should be considered after 2–3 years.

Spontaneous recovery of AIH may occur and it is difficult to justify treating patients with normal serum transaminases and globulin/IgG and with minimal necroinflammatory activity on liver biopsy. A previous history of spontaneously resolving hepatitis is found in 20% of patients re-presenting with AIH.162 Such apparently resolving patients should be followed up because severe AIH relapse may occur. Drug-related AIH may resolve on drug withdrawal, although this is poorly documented because most reported cases have received immunosuppressive treatment.56

C3. Standard regimes

Standard induction treatment for patients with moderate or severe AIH (detailed in figure 1) has been firmly established by controlled trials published in the early 1970s and recently reviewed.2 Two trials from London163–165 demonstrated the efficacy of prednisolone 15–20 mg/day over placebo and over azathioprine alone in improving serum liver tests and globulin and, more importantly, 2–4-year survival. A contemporaneous study from the Mayo Clinic166 demonstrated the superiority of two regimes: prednisolone alone (starting at 60 mg/day, reducing to 20 mg/day over 4 weeks) and prednisolone (in half this dose) combined with azathioprine 50 mg/day over placebo and over azathioprine alone with regard to improvement in liver tests, liver histology and survival. These two ‘active’ regimes had similar beneficial effects but the combination regime was associated with fewer side effects than prednisolone alone (10% vs 44%). In patients receiving placebo, mortality exceeded 50%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Suggested induction strategy for autoimmune hepatitis (AIH). ALT, alanine aminotransferase; AST, aspartate aminotransferase; TPMT, thiopurine methyltransferase.

In the Mayo Clinic study, 80% of patients on either prednisolone alone or prednisolone plus azathioprine combination therapy achieved a serum ALT of less than twice the upper limit of normal within 6 months. Histological remission (loss of interface hepatitis on 6-monthly liver biopsy) lagged behind clinical and biochemical remission by several months, but was achieved in 75% of patients after 18 months of active treatment (in contrast to only 20% of those given azathioprine alone or placebo).

In a follow-up controlled study167 the Mayo Clinic group evaluated a fifth treatment strategy in AIH—namely, prednisolone alone, again starting at 60 mg/day but tapered and titrated to maintain serum transaminases at less than twice the upper limit of normal. This regime improved clinical symptoms and serum liver enzymes as well as the other two active regimes and, because the mean maintenance dose of prednisolone was only 10 mg/day, side effects were less severe than the initial prednisolone regime (maintaining dose at 20 mg). However, histological remission was achieved after 24 and 36 months treatment in only 4/16 and 3/9 cases on the titrated prednisolone regime compared with 13/19 and 11/14 of patients receiving prednisolone 20 mg/day and 16/20 and 16/17 receiving the predisolone/azathioprine combination regime. The significance of continuing activity on follow-up liver biopsy is discussed below. Unfortunately, the combination of titrated dose prednisolone and azathioprine has not been evaluated.

HBV_sAg positivity was found in 4% and 14% of subjects in two of these studies165 166 and was not tested in the third.163 Inevitably, none of the patients in these studies had HCV excluded. However, the high prevalence of autoantibodies and the raised serum globulins in nearly all the patients in the earlier trials suggest that the vast majority indeed had AIH. Furthermore, only about 5% of patients previously considered to have type 1 AIH were positive for HCV.168

These studies therefore suggest that, of the regimes evaluated, the best one for most patients (combining maximum efficacy with minimal side effects) is the prednisolone/azathioprine combination regime (figure 1). Until recently no other treatment strategy had been compared with and been shown to be superior or equivalent to this regime. In a recent multicentre randomised controlled trial,169 patients with AIH and without cirrhosis given budesonide 9 mg/day plus azathioprine 1–2 mg/kg/day for 6 months achieved normalisation of serum transaminases more quickly and had fewer side effects than those given prednisolone plus azathioprine (see also section F). In contrast to the early trials in AIH, viral hepatitis was rigorously excluded. However, no follow-up histology data were presented and the blinded phase of the trial lasted only 6 months. Thus, more long-term results are required for budesonide and, currently, the recommended initial treatment for most patients remains prednisolone plus azathioprine. However, budesonide 9 mg/day plus azathioprine may be considered in non-cirrhotic patients with severe (actual or anticipated) steroid-related side effects such as psychosis, poorly controlled diabetes or osteoporosis (see section C4).

Several variants of the initial prednisolone plus azathioprine regime have been proposed:

1. Prednisolone is sometimes started in a higher dose than 30 mg/day (with azathioprine). This dose is recommended in the AASLD guidelines,4 however up to 1 mg/kg/day73 170 plus azathioprine has been proposed recently. The prednisolone is then reduced gradually to 10 mg/day over 2–3 months as serum transaminases fall. Such a strategy is likely to cause more steroid-related side effects and this may be problematic in frail elderly patients. However, in non-cirrhotic patients it may result in more rapid normalisation of serum transaminases (77% after 6 months in one preliminary report170), in contrast to only 39% with standard dose prednisolone in a recent randomised trial.169 However, while prolonged failure of transaminases to normalise171–173 and absence of an early fall68 174 are both associated with worse outcomes (see below and also table 6), the rate of normalisation of transaminases is not adequately established as a prognostic factor to be a ‘hard’ treatment endpoint. This strategy therefore needs a firmer evidence base.

2. Azathioprine is often started a few weeks after the prednisolone. Although not evidence-based, this may be reasonable, especially if there is diagnostic doubt. A rapid improvement of liver tests in response to prednisolone may be reassuring in practice in relation to a diagnosis of AIH, although such a response does not in itself increase the likelihood of AIH by the revised IAIHG criteria (see section C2). Also, institution of azathioprine may be delayed pending results of thiopurine methyltransferase (TPMT) assay (see section C4).

3. High-dose ‘pulse’ prednisolone (90 mg every 5 days) without azathioprine was evaluated in the hope of reducing steroid-associated side effects but was shown to be inferior to the combination regime in inducing remission.186

4. Although in the Mayo Clinic trials azathioprine was used in a dose of 50 mg/day, in Europe it is typically started at the (usually) higher dose of 1 mg/kg/day. In some centres prednisolone plus azathioprine in a still higher dose (up to 2 mg/kg/day) is used as initial treatment. The recent AASLD guidelines4 recommend prednisolone plus azathioprine 1–2 mg/kg/day. However, use of azathioprine in initial doses exceeding 1 mg/kg/day is not evidence-based.

5. The combination of azathioprine 1 mg/kg and titrated dose prednisolone (to maintain normal ALT) may be as effective as the standard regime (and have fewer side effects) but has not been evaluated.

C4. Side effects

In the above trials163 165 166 side effects with the prednisolone-only regimes were problematic. Cushingoid features developed in 20–50% of patients, diabetes in 15–20% and psychosis, hypertension, cataracts and osteoporotic vertebral collapse in 5–10%. These were less common (5%) on the combination regime. However, subsequent studies of patients on this regime187 188 suggest about a 30% prevalence of steroid-related side effects including, most notably, weight gain which improves on stopping the prednisolone. In patients without cirrhosis in whom severe side effects such as psychosis, difficult-to-control diabetes or severe osteoporosis preclude the use or continued use of prednisolone, azathioprine alone is ineffective.166 In these patients, budesonide 9 mg/day plus azathioprine 1 mg/kg/day may be considered based on the results of a recent controlled trial (see section F).169 However, osteoporosis should largely be preventable by routine prescription of calcium and vitamin D and by selective use of bisphosphonate therapy (see sections C5 and E).

About 25% of patients with AIH develop side effects on azathioprine, requiring withdrawal of the drug in about 10% of cases.19 64 188–190 Side effects are more common in patients with cirrhosis.189 About 5% of patients develop a severe early reaction with fever, arthralgia, a skin rash and influenza-like symptoms.191 Approximately 10–20% of patients develop nausea and anorexia, which may improve despite continuation but may require dose reduction or withdrawal. Rarer side effects include skin rash, pancreatitis192 and cholestatic hepatitis.193

The most serious side effect of azathioprine is marrow depression, usually manifest as a fall in white cell and neutrophil count which require regular monitoring (see section C5). The risk of marrow toxicity can sometimes be predicted by genotyping for or measuring activity of the enzyme thiopurine methyltransferase (TPMT), which catalyses conversion of 6-mercaptopurine (the active metabolite of azathioprine) into inactive products.194 About one in 300 people is homozygous for a low-activity allele of the gene and has very low enzyme activity. In these patients active 6-mercaptopurine metabolites accumulate and serious toxicity is common, although it may be avoided by use of low doses with careful monitoring of metabolites in the blood.195 Heterozygosity for the low-activity allele with intermediate enzyme activity is found in about 10% of people. Studies in patients with AIH suggest that neither heterozygosity for the low-activity allele, modest reductions in TPMT activity nor 6-mercaptopurine metabolite levels are reliably predictive of azathioprine efficacy or toxicity.189 190 196 However, TPMT measurement should be considered to exclude homozygous TPMT deficiency, and this is recommended in patients with pre-existing leucopenia (usually due to hypersplenism).

In patients with severe TPMT deficiency and in those intolerant of azathioprine, the prednisolone-only regime or a lower dose of prednisolone combined with mycophenolate (see section F4) may be used. The prednisolone-only regime may also be used in patients with a recent history of cancer because of a potential (though unproven) link between azathioprine and malignancy (see section D1). It may also be preferred in those with severe cytopenia due to hypersplenism or coincidental blood diseases. Moderate leucopenia, common in cirrhosis, probably does not increase the risk of azathioprine-related marrow depression per se but is likely to complicate haematological monitoring.

C5. Monitoring and additional management

Patients should be asked about and/or tested for immunity to hepatitis A and hepatitis B infection and susceptible patients should be offered vaccination as soon as possible.

Patients on combination therapy should have baseline and weekly on-treatment monitoring of liver tests, blood sugar and blood count for 4 weeks and then 1–3 monthly thereafter, depending on the responses. All should receive calcium and vitamin D supplementation. DEXA bone density scans should be performed at commencement of prednisolone-containing treatments and repeated at 1–2-yearly intervals while prednisolone treatment is continued (see section E). Screening for glaucoma and cataracts should also be considered after 12 months prednisolone treatment.

C6. Endpoints of initial treatment

In 80–90% of patients with moderate/severe AIH, serum ALT falls after starting treatment. Usually the fall commences within 2 weeks. As transaminases fall, clinical symptoms resolve and liver function (albumin, bilirubin and prothrombin time) shows marked improvement within 3–6 months of starting prednisolone with or without azathioprine. Although ascites and encephalopathy may also resolve, such patients should be discussed with a transplant centre.

In 5–10% of patients liver tests do not improve. Such patients tend to be younger,197 to have an acute presentation with severe jaundice and a high model for end stage liver disease (MELD) score198 which shows little change over 7 days of corticosteroid therapy.199 In a recent report, only one of 12 patients with AIH presenting as fulminant liver failure improved with corticosteroid treatment and 10 required liver transplantation.72 Failure to respond may be associated with confluent necrosis on biopsy.151 Non-response may be more common in non-Caucasian patients.13

In non-responding or very slowly responding patients without liver failure, prednisolone may be increased to 60 mg/day and azathioprine to 2 mg/kg/day if tolerated.200 The possibilities of non-compliance and of malabsorption should be considered, as should admission to hospital for intravenous hydrocortisone or methylprednisolone. Of alternative drugs (see section F), tacrolimus may be the most useful199 but patients are sometimes resistant to all medications. Referral to or discussion with a physician with expertise in treating AIH should be considered.

If there is confluent necrosis on the biopsy, evidence of liver failure or if the serum bilirubin and MELD score do not improve rapidly in patients with jaundice, the chances of mortality are high71 72 151 197–199 and early referral to a liver transplant centre is recommended.

As the serum ALT falls, the initial dose of prednisolone should be reduced to 10 mg/day, usually by 5 mg/day every week. Biochemical remission in the initial Mayo Clinic trials and in many subsequent studies was defined as serum AST less than twice the upper limit of normal and was achieved by 80% of patients on prednisolone-based regimes. In most patients serum transaminases eventually fall to within the normal range,155 171 177 182 although this may take 12 months or more. Somewhat confusingly, the IAIHG adopted two definitions of remission—a fall to less than twice normal and full normalisation of serum AST.103 However, it is now generally agreed (and explicit in the AASLD guidelines)4 that complete normalisation of transaminases should be the aim. This is because even mild persisting elevations are predictive of persisting hepatitis,201 of relapse following treatment withdrawal,178 of progression to cirrhosis173 174 and of a poor outcome.155 171–173

Even after serum transaminases normalise, treatment should be continued. In the Mayo Clinic trials166 167 prednisolone was maintained at 10 mg/day together with azathioprine. Patients on this regime had higher histological remission rates after 2–3 years than those treated with prednisolone alone in a dose titrated to maintain serum transaminases at less than twice normal.167 Histological remission lagged behind biochemical remission by several months.166 This might justify a continuing dose of 10 mg/day prednisolone if tolerated. However, the AASLD guidelines suggest continuing at a lower dose of 5–10 mg/day (depending on tolerance) together with azathioprine (50 mg or 1 mg/kg/day) for a total of at least 2 years.4

Although there is no consensus on optimal duration,3 177 it is recommended that treatment be continued for long enough to make histological resolution ‘likely’, based on the initial Mayo Clinic trials.166 167 In practice this means for 2–3 years, with normal transaminases for at least 18 months.

Histological remission (absence of interface hepatitis) has been an important aim of treatment.166 167 187 188 It lags behind biochemical resolution and is achieved in only 15–35% of patients after 12 months on prednisolone-based regimes. In the two Mayo Clinic controlled trials this rose to 60% and 78% after 24 months and 60% and 87% after 36 months of treatment.166 167 In a further study from the Mayo Clinic, 88 of 115 patients (77%) achieved histological remission.179

Thus, some patients still have interface hepatitis despite normalisation of transaminaseses. The proportion was 55% in one study,201 but this included many biopsies taken relatively early in the course of treatment. After treatment was stopped, 20% of patients with normal serum AST and 50% of those with AST 1–2 times normal had interface hepatitis.201 Serum globulin did not add to the predictive value of the serum AST. Recently, normalisation of serum IgG has been suggested as being predictive of histological remission together with transaminases,202 but this requires further evaluation.

Reasons for obtaining a liver biopsy about 2 years after liver tests have normalised to confirm histological remission include: (a) regimes to prevent relapse of AIH (see section D1) have been evaluated specifically in patients with documented histological remission; (b) progression of fibrosis on serial liver biopsies occurs in 20–40% of patients with AIH and is positively correlated with the degree of residual histological inflammation183 202 203 so fibrosis tends to regress in patients with minimal inflammation on follow-up biopsy and to progress in those with persisting inflammation (indeed, cirrhosis develops de novo in 10–50% of patients with AIH while on treatment, see section D2); (c) follow-up biopsy is partially predictive of AIH relapse following treatment withdrawal, so a normal repeat liver biopsy179 and absence of portal tract plasma cells139 are associated with a lower risk of relapse (see table 6); (d) preliminary evidence suggests that histological resolution may be independently predictive of long-term survival, with an optimal outcome associated with follow-up Ishak necroinflammatory grade ≤3.203 204

However, routine follow-up biopsy has not yet been shown to affect ultimate patient outcome and is often not performed. Rather, the aim is sustained normalisation of serum transaminases.4 171 177 The decision to perform a repeat biopsy must therefore be individualised. Factors weighing against it include an elderly or reluctant patient or one in whom liver tests had not fully normalised. The positive predictive value of a raised serum AST for active AIH is high (80% if 1–2 times the upper limit of normal and 100% if increased more than twofold).201 This justifies delaying the repeat biopsy until the liver tests have normalised over at least 12 months. However, a repeat biopsy should be considered if azathioprine hepatotoxicity is suspected; this is in part dose-dependent and has recently been associated with high blood levels of 6-methylmercaptopurine metabolites.205 206

If there is clinical and histological remission, the prednisolone should be reduced gradually. A suggested regime is reduction of 2.5 mg/day each month with monitoring of liver tests. Whether azathioprine is continued or not depends on the long-term management strategy (see section D3).

If either serum liver enzymes improve but remain abnormal or if, despite normal liver enzymes, the repeat liver biopsy continues to show interface hepatitis, the optimum strategy is unclear. The possibility of non-compliance should be considered, as should discussion with a centre experienced in the management of AIH. Increasing the dose of prednisolone to >10 mg/day is an unattractive option. Increasing the dose of azathioprine to 2 mg/kg/day (the dose to prevent relapse, see section D1) together with 5–10 mg/day prednisolone may be considered. Alternatively, other agents (see section F) may be tried. Of these, mycophenolate appears of limited efficacy in patients not responding to azathioprine.207 Tacrolimus and ciclosporin may be effective. However, more data are needed on these agents. Whatever regime is used, a repeat liver biopsy should be considered after a further 12–18 months. Complete biochemical and histological resolution is the aim; however, this may not be achievable in some patients. The aspiration should be the lowest achievable histological and biochemical activity with a minimum of side effects.

D1. Relapse

Within 12 months of stopping treatment following biochemical and histological remission, 50–90% of patients have a disease relapse (defined by IAIHG criteria as serum ALT >3 times the upper limit of normal).172 208–211 Occasionally, relapses occur >10 years after treatment withdrawal.212 Although other causes of acute liver injury (viral, drugs) should be considered, a liver biopsy is not usually necessary to confirm the presence of AIH relapse because of the high predictive value of AST >2 times the upper limit of normal.201 With mild elevations in well patients, liver tests should be repeated after 1–2 weeks before assuming that relapse has occurred.

In retrospective analyses, relapse of AIH has been associated with several parameters (table 6). Relapse is more likely in patients who have been slow in achieving biochemical remission and in those with continuing active inflammation prior to treatment withdrawal, as evidenced by persistent elevation of serum transaminases and/or serum globulins and IgG178 and by persistence of plasma cells in portal tracts on liver biopsy. In contrast, only 30% of patients with complete resolution of AIH on follow-up liver biopsy relapse.179 Relapse of AIH following treatment was associated with a shorter length of initial treatment in one retrospective study177 but not in another.178 Few of these reported associations have been independently confirmed.213 Relapse also appears to be uncommon when there is an identifiable precipitant for the initial presentation such as a drug.56

Following reintroduction of the initial treatment regime, more than 80% of patients again achieve biochemical remission, usually within a few months. There are no data on the rate of histological remission. However, the increased doses of prednisolone required to reinduce remission may have further side effects. Patients with multiple relapses of AIH were more likely to have treatment-related side effects than patients who were maintained in remission.210 211 In some studies they were more likely to develop progressive fibrosis or cirrhosis155 184 211 and to either die of liver disease or require transplantation.155 184

Maintenance regimes have therefore been evaluated to try and prevent relapse of AIH. These studies have recently been reviewed.2 All have been in patients with documented histological remission. The relapse rate after 1 year was only 8% when standard combination treatment was continued, but was 32% when azathioprine was withdrawn and prednisolone continued.214 In a follow-up trial187 standard combination treatment was continued in one group and, in the other, prednisolone was phased out gradually and azathioprine continued, with the dose increased from 1 to 2 mg/kg/day. The main clinical advantage of this regime was disappearance of steroid-related side effects, although temporary arthralgia after prednisolone withdrawal was problematic in over half of the patients After 1 year no patient developed clinical or biochemical relapse, although two patients had recurrent inflammation on routine liver biopsy. In a follow-up study of 72 patients maintained on azathioprine 2 mg/kg/day,188 83% of patients remained in clinical and biochemical remission over a median follow-up period of 67 months. Note that these patients had already been in histological remission for at least 12 months and an additional 29 patients were excluded from the study because they did not meet this entry criterion.

The main disadvantages of long-term maintenance of azathioprine are the need for continued blood count monitoring (although the risk of serious marrow depression decreases with time) and the theoretical risk of cancer. An increased risk of malignancy has been reported in patients following renal transplantation215 and with rheumatoid arthritis treated with azathioprine,216 but a cause and effect relationship with azathioprine was not proven. The results of three studies are consistent with a slight increase in the risk of cancer following a diagnosis of AIH (RR 1.34–1.51), although the increase achieved statistical significance in only one.217–219 The risk was not related to dose and duration of azathioprine treatment. However, although unproven, an increased risk of malignancy remains possible and might limit enthusiasm for lifelong azathioprine treatment. It may be prudent to advice patients on long term azathioprine to avoid excessive exposure to sunlight.

Although reduction or withdrawal of azathioprine is a reasonable strategy in a patient who has been free of AIH recurrence over several years on the maintenance regime, this strategy has not been formally evaluated. In 26 patients in one follow-up study,188 the dose of azathioprine was electively reduced to 1 mg/kg/day; five of these patients subsequently relapsed. In a recent report19 azathioprine was withdrawn completely after 2–12 years (median 5) in 22 patients. Nine patients remained in remission for 3–12 years (median 7) and 11 relapsed. Following retreatment of the relapse, patients are often maintained on azathioprine and small doses of prednisolone,19 although the added value of prednisolone in preventing further relapses has not been assessed.

An alternative approach, more commonly used in North America than in Europe, is not to use routine maintenance azathioprine treatment but to stop treatment once remission is attained and to then treat relapses as they occur.64 171 210 211 About 25–36% of patients achieve a ‘sustained remission’, defined as serum AST persistently <3 times the upper limit of normal, a looser definition than that of biochemical remission following initial treatment.210 211

A low-dose prednisolone maintenance regime has also been suggested for patients with multiple AIH relapses.220 In 22 patients (seven of them also on azathioprine 50 mg/day), the dose of prednisolone was reduced to maintain serum AST at <5 times the normal value. The median maintenance dose of prednisolone was 7.5 mg/day and the main advantage of this strategy was a reduction in the severity of side effects to steroids compared with a group of 31 patients with multiple relapses who received repeated episodes of standard therapy for each relapse. However, although follow-up histology data were not presented, it is doubtful that the AIH was fully suppressed in these patients. Progression to cirrhosis was common (55% overall) and, of the 22 patients on the low-dose regime, two died of liver disease and two required transplantation. The study further illustrates the relatively poor outlook in patients with serially relapsing AIH, and other treatments should be considered in these patients (see section F).

Consideration of the merits of these long-term strategies (expectant management of relapses, maintenance azathioprine and low-dose prednisolone) must be informed by a discussion of disease progression.

D2. Progression of disease

Suggestions that the outlook for patients with treated AIH is not different from that of the general population have been based on mean follow-up periods of about 10 years and maximum follow-up of only about 20 years. They may be over-optimistic in the long term. In some studies 10-year survival is about 90% and not different from the general population.180 221 However, others have reported lower survival rates,18 64 especially after longer periods of follow-up.61 155 The standardised mortality ratio has been reported as above unity: 3.7 in a study from the Danish National Registry222 and 1.63 in a recent study of 245 UK patients followed up for a median of 9.4 years.155

Complementary to the main long-term management goal in AIH of preventing liver-related death or need for transplantation is prevention of fibrosis progression. Studies involving serial liver biopsies19 223 suggest that fibrosis improves in about half of patients with treatment. However, in about 25% of cases fibrosis progresses despite treatment, and this is associated with failure to suppress inflammation.183

Cirrhosis develops during follow-up in 30–50% of patients,61 155 180 181 184 although lower rates have also been reported.68 172 224 The development of cirrhosis is associated with several parameters of suboptimal treatment response and also with increased liver-related mortality or transplantation (table 6). Cirrhosis at presentation occurs in about 30% of patients and is associated in some61 64 159 but not all172 180 studies with a poorer outlook than in patients without cirrhosis.

Associations and predictors of progressive fibrosis, development of cirrhosis and liver-related death or transplantation have been demonstrated in retrospective analyses (table 6). However, it is not known if the long-term treatment strategy (see section D1) influences progression of disease. In the King's College follow-up study188 of long-term maintenance azathioprine 2 mg/kg/day, only one of 73 patients died of liver failure. However, in recent studies19 155 of patients, most of whom received maintenance azathioprine, 14–20% eventually died of liver disease or required liver transplantation. This outlook is not obviously different from that observed in studies in which maintenance azathioprine was not used.64 171 172 180 Unfortunately these two long-term strategies have not been formally compared with regard to these ‘hard’ endpoints. Of concern, it has not been shown that any regime prevents development of cirrhosis.

Hepatocelluar carcinoma was previously regarded as a rare complication of AIH.225 However, recent reports suggest a rate of 4–6% overall and 10–20% in patients with cirrhosis.218 226 227 Therefore, although not common practice, screening for hepatocellular carcinoma with 6-monthly serum α-feta protein measurements and liver ultrasound may be considered in otherwise healthy patients with AIH and cirrhosis (both men and women) because of the 10–20% risk over 20 years.

D3. Choice of long-term maintenance strategy in patients who have attained remission

The decision as to which strategy to pursue in patients who have attained remission must be individualised. Features favouring withdrawal of treatment and delaying reinstitution of maintenance azathioprine therapy until after the first relapse would include: (a) absence of cirrhosis or decompensation; (b) absence of the features in table 6 associated with relapse; (c) good tolerance of initial prednisolone treatment; (d) a potential precipitant of the initial episode of AIH such as a drug or documented viral infection; and (e) a history of malignancy. On the other hand, continuing azathioprine long-term as a maintenance strategy is recommended in younger patients and in those with predictors of relapse including LKM- and SLA-positive patients (table 6), in those with observed or anticipated prednisolone-related side effects (including osteopenia, see section E) and in those with cirrhosis or decompensation. After treatment of one relapse, continuation of azathioprine 2 mg/kg/day as long-term maintenance therapy is recommended. Whatever strategy is pursued, monitoring should be lifelong.

F1 Ciclosporin A (CyA)

In patients with AIH not responding to standard therapy, CyA has been shown to be of clinical benefit although relapses occurred if the dose of CyA was reduced.232–234 In an open-label trial, 19 patients (9 treatment-naïve) treated with CyA showed reductions in serum aminotransferases and histological activity index scores over 6 months.235 There was no significant effect on serum creatinine. In a multicentre study of 32 children, CyA was administered as monotherapy for 6 months (target trough levels 200–250 ng/ml), followed by low doses of prednisolone and azathioprine which were given for 1 month, after which CyA was stopped.236 Two patients were withdrawn from the study and, of the remaining 30 patients, ALT normalised in 25 by 6 months and in all by 1 year. Adverse effects were mild and reversible on withdrawal of treatment. In a study of 84 children recruited from five centres between 1994 and 2001, CyA was administered during the first 6 months in a protocol similar to that described by Alvarez et al236 and, after 6 months for patients with AST/ALT levels lower than twice the upper limit of normal, standard therapy was initiated.237 Normal aminotransferase levels were observed in 94% of patients, with 72% of normal results achieved within the first 6 months of treatment. Higher bilirubin levels and the presence of portal hypertension at diagnosis predicted a delay in achieving remission.237 Adverse effects related to CyA appeared mild and transient, with standard therapy not implicated in disease relapse during follow-up.

In other series CyA has been used predominantly as a salvage strategy or in the context of relapsing or non-responsive AIH. Results in these situations have been favourable, although no long-term reports exist to evaluate safety.233 238 239 Therefore, in considering initiation of CyA, its toxicity profile including the long-term risks of hypertension, renal insufficiency, hyperlipidaemia, hirsutism, infection and malignancy must be balanced against its potential benefits.

F2. Tacrolimus

Tacrolimus (FK 506) is a macrolide antibiotic with 10–200 times greater immunosuppressive potency than CyA.212 Its mechanism of action is similar to that of CyA, although it binds to FK binding protein (an alternative immunophilin). In an open-label preliminary trial in 21 patients with AIH treated with tacrolimus (at trough drug levels of 0.6–1.0 ng/ml), biochemical improvement was demonstrated after 3 months.240 Serum urea and creatinine levels were raised within a year of treatment.241 Tacrolimus has been used predominantly as salvage therapy in AIH in relatively small series or in case reports.241–245 Recently, tacrolimus was successful in seven of nine patients with acute AIH who did not respond to corticosteroids.199 For most patients remission can be achieved with tacrolimus, either alone or in conjunction with corticosteroids, although the limitation of all series relates to the short degree of follow-up.

F3. Budesonide and deflazacort

Budesonide is a second-generation corticosteroid with 15 times the affinity of prednisolone for the glucocorticoid receptor. When taken orally it has a 90% first-pass metabolism in the liver, although in patients with cirrhosis with shunts, the metabolism may be variable.246 247 In one study248 13 patients (11 of whom were intolerant of standard therapy) were treated with 6–8 mg/day budesonide, reduced to 2–6 mg/day after 6–10 weeks. Both ALT and serum IgG improved and there were no reported adverse steroid effects. In a second study,249 10 patients who required continuous treatment to prevent disease exacerbation were treated with budesonide 9 mg/day with only three patients having a sustained effect. In an uncontrolled study, however, budesonide controlled disease activity in 7/9 patients with refractory or prednisolone-dependent disease without excessive side effects.250 In untreated patients a dose of 9 mg/day was used to induce remission and, in this group, 7/12 patients (58%) reached complete remission and 3/12 (25%) had a partial response, with treatment being well tolerated in 10/12 cases (83.3%).251 Information is not available on the long-term outcome in patients treated with budesonide.

In a recent multicentre randomised controlled trial in patients with AIH without cirrhosis,169 budesonide 9 mg/day plus azathioprine 1–2 mg/kg/day for 6 months was more effective in achieving normalisation of serum transaminases and produced fewer steroid-related side effects than prednisolone plus azathioprine. Given the short trial duration and the fact that no follow-up histology data were presented, routine use of this regime in treatment-naïve patients is not currently recommended. However, its use should be considered in non-cirrhotic patients who are intolerant of prednisolone.

In an open-label trial, 15 patients with AIH maintained in remission on prednisolone were converted to deflazacort 7.5 mg/day.252 ALT and IgG levels were raised minimally in most patients, although there were no apparent ill effects and remission was sustained during 2 years of follow-up.

F4. Mycophenolate mofetil

This is the prodrug of mycophenolic acid (MMF) which blocks de novo purine synthesis. MMF has been used predominantly in patients with refractory AIH or with azathioprine intolerance.207 244 253–256 Most studies have used 2 g/day in divided doses, initially with corticosteroids but with the aim of reducing these. In one study the dose of prednisolone was decreased from a median of 20 mg/day to 2 mg/day after 9 months, with histological improvement noted in all patients.253 In a further case series 5–15 patients unresponsive to or intolerant of standard therapy were given MMF, with improvement in biochemical and histological indices.254 257 In larger series (29–36 patients) up to one-third of patients discontinued the drug due to poor tolerance or side effects.207 256 258 Only half of patients attained biochemical remission and follow-up histology data were not reported. Patients in whom azathioprine had been ineffective seemed to have a poorer response to MMF than those who had been azathioprine-intolerant.207 258 In a more promising recent report of 59 treatment-naïve patients,259 88% had a complete biochemical response to MMF and prednisolone. However, further data are needed, especially on efficacy in inducing histological remission, before MMF can be recommended as a first-line treatment for AIH.

There are very few data on the long-term safety of MMF.260 Development of histological changes were noted in a patient with AIH, including cytoplasmic features of adaptation and nuclear alterations within hepatocytes,261 and there have been sporadic reports of cerebral lymphoma in patients who have received MMF for other autoimmune diseases.262–264 MMF is contraindicated in pregnancy.

F5. Other agents with acectdotal evidence of efficacy

In a Japanese study eight patients received ursodeoxycholic acid (UDCA) 600 mg/day for 2 years in whom a significant biochemical improvement was demonstrated.101 In four patients a biopsy after 12 months showed improvement in inflammation. However, in another study 37 patients refractory to corticosteroid therapy were randomised to UDCA or placebo for 6 months, together with their standard treatment regimen.265 No significant benefit from UDCA was found.

Maintenance therapy with cyclophosphamide for up to 12 years without relapse or serious side effects has been successfully achieved in three patients.266 In case reports, methotrexate,267 infliximab268 and rituximab269 given to patients with resistant AIH resulted in sustained biochemical remission and histological improvement.

G1. Management of AIH in pregnancy107270–272

Of 162 women with AIH attending King's College Hospital between 1983 and 1998, 18 (7 with cirrhosis) had 35 pregnancies which resulted in 31 live births.270 Birth abnormalities were seen in only two cases. At conception, 15 patients were receiving azathioprine (9 with prednisolone). Azathioprine was continued in 10 patients. Flares in disease activity occurred during four pregnancies and within 3 months of delivery in a further four. In another series of 42 pregnancies in women with AIH there were 11 adverse outcomes and four serious maternal complications.271 The unexplained adverse outcomes were associated with the presence of antibodies to anti-SLA/LP and anti-Ro/SSA; 21% of the patients had flares during pregnancy and 52% of patients had post-partum flares. In a survey of 63 pregnancies in patients with AIH a higher rate of cesarian section was observed but there was no increase in stillbirth or fetal malformation rate compared with normal controls.273

In none of these series was any relationship apparent between azathioprine use during pregnancy and an adverse outcome. Larger studies of patients with inflammatory bowel disease274 also support the relative safety of azathioprine/6-mercaptopurine during pregnancy. Continuation of this drug during pregnancy may therefore be justified. Indeed, in another series of 14 patients with AIH, immunosuppression was reduced during the second trimester.275 Following delivery (or stillbirth in one patient), 12 of the 14 patients had a rapid flare of AIH. AIH presenting de novo following delivery has also been reported.271 Taken together, these data provide some support for a strategy of minimal adjustment to prednisolone/azathioprine-based immunosuppressive regimes during the course of pregnancy so that the risk of flare during pregnancy and post-partum can be minimised. Similar considerations may apply to ciclosporin and tacrolimus276 based on experience in patients following liver transplantation. However, the decision to continue or to stop immunosuppression before or after conception should be individualised and should always involve full discussion with the patient. Mycophenolate is potentially teratogenic and is not recommended during pregnancy.

G2. Liver transplantation

About 10–20% of patients with AIH will require liver transplantation during their lifetime. There are two distinct indications. The first is severe acute AIH resulting in acute or subacute liver failure. The critical importance of early referral to or discussion of these patients with a transplant centre has been previously discussed (sections B1 and C6). The second and more common indication for liver transplantation is decompensated chronic liver disease and/or hepatocellular carcinoma, often in a patient with longstanding AIH. Table 6 (right hand column) lists the features associated with and predictive of this outcome. The same indications for liver transplantation apply as for other aetiologies of cirrhosis in so far as the development of ascites, variceal bleeding, hepatic encephalopathy, spontaneous bacterial peritonitis or hepatorenal syndrome impact significantly on survival. These include a MELD score of >15 or a Child-Pugh score of >10.277

However, if possible, patients should be referred for consideration of liver transplantation before they develop end stage liver disease. Pointers to impending liver decompensation include a variceal bleed, ultrasound showing a small ‘fibrotic’ liver, a falling serum albumin and development of even mild ascites or ankle oedema.

Five-year survival following transplantation for AIH is about 75%.278–282 In a recent Europe-wide experience it seemed to be worse than in patients transplanted for PBC.282

Recurrence of AIH, originally described in 1984,283 is seen in about 20% of recipients.284–286 Diagnosis of recurrent AIH is limited by the absence of a specific marker. Autoantibodies such as ANA or ASMA tend to disappear after transplantation but may reappear, albeit at lower titres than before transplantation.287 288 The features of recurrent AIH in the liver graft are similar to those of AIH before transplantation. The most frequent overlapping diagnosis is acute cellular rejection, and both AIH and acute rejection entities share common biochemical and histological features and both respond to corticosteroids.288 Indeed, patients with AIH hepatitis have an increased risk of acute cellular rejection.286 288 289 In addition, recurrent AIH may precipitate chronic rejection.286 A review by the Banff Working Group has recommended diagnostic criteria for recurrent AIH after transplantation.290 Recurrent AIH is usually managed by either maintenance of corticosteroids long term or by continuation of azathioprine in the immunosuppression regimen.284

‘De novo’ AIH following liver transplantation

This occurs in association with serological and histological features compatible with AIH in patients who have undergone transplantation for aetiologies other than AIH.291–293 It has been classified by the Banff Working Group as requiring the same criteria as recurrent AIH after transplantation.290 In some patients retransplantation was required because of severe graft dysfunction.293 Two patterns of disease were described, one associated with detectable anti-LKM1 antibodies at high titre in conjunction with a serum AST level >500 IU/l and a second associated with the presence of ANA or ASMA at lower titres (>1:80) and lower AST levels.293 Treatment has usually involved the reintroduction of corticosteroids and azathioprine.291 293 Subsequent descriptions of this type of graft dysfunction have been controversial, as has the terminology associated with the condition. The terms ‘de novo AIH’, ‘alloimmune hepatitis’ or ‘graft dysfunction mimicking AIH’ have been proposed.293–295

A phenomenon of immune-mediated hepatitis has also been reported in patients who have undergone treatment for recurrent HCV infections following liver transplantation.33 35 In these reports an aggressive plasma cell predominant hepatitis has been described and treatment with steroids and/or azathioprine resulted in prompt and dramatic response of liver enzyme abnormalities, although relapse occurred following discontinuation of the drug.35

G3. Overlap syndromes