EASL Clinical Practice Guidelines

Difficult to treat patients

Most, but not all patients respond well to conventional treatment, and those who do respond may develop side effects related to the treatment. Depending on the remission criteria used (full response with normal aminotransferase activity, normal immunoglobulin concentration and normal histology or only biochemical normalisation), at least 10–15% of patients seem to be refractory to standard treatment, as a result of non-compliance, partial compliance or true non-response. Furthermore, some patients might have variant syndromes with features of PSC or PBC precluding full normalisation of liver enzymes. Finally, co-morbidities may limit therapeutic options and thus alter the management.

Biochemical response to standard immunosuppression is the rule and is often viewed as an additional diagnostic criterion. As a result, non-response should question the diagnosis and adherence to treatment. Non-response is not well-defined in AIH. A lack of a reduction of transaminase by more than 25% after two weeks should be regarded as non-responsive. Numerous diseases can resemble AIH including Wilson’s disease, NASH, DILI and atypical forms of PSC or PBC (variant syndromes). These conditions may be unrecognised at presentation and should be reconsidered if apparent non-response is observed. Moreover, AIH may undergo transitions during its course, with a cholestatic syndrome emerging that might be refractory to the original treatment (secondary non-response). Lastly, another condition may be superimposed upon the original process during the course of AIH such as viral infection, drug toxicity or fatty liver.


Different grades of non-response have to be considered; null response (treatment failure) with or without immediate severity and incomplete (partial) response requiring individualised therapeutic management. In patients with sub-optimal response despite reconfirmation of diagnosis and adherence, dosage of prednisolone and azathioprine should be increased or alternative medications should be used.

Treatment failure

A) With immediate severityA particularly challenging scenario is the setting of severe acute presentation. Indeed, treatment failure is more likely to be seen in patients presenting (sub) fulminant disease. Unfortunately, there is a paucity of published data on patients with acute severe AIH at presentation; consisting mostly of anecdotal case reports or small case series with varying inclusion criteria [[42], [206], [272], [273]]. As a result, it remains unclear whether such patients should be given a trial of corticosteroids, be priority listed for LT, or both; and if corticosteroids are indeed initiated, how and at what time point failure of medical treatment should be defined [274]. Nevertheless, prognosis is poor with overall mortality ranging from 19% to 45% and rate of LT required range from 9% up to 81%. The largest study came from the UK and included 32 patients with acute severe AIH as an acute presentation with an INR of ⩾1.5 at any time without histological evidence of cirrhosis [272]. Twenty-three patients were treated with corticosteroids (⩽40 mg/day) of whom ten (48%) required LT, while all nine untreated patients required LT (p = 0.01). Untreated patients demonstrated higher MELD scores at presentation and a non-significant decrease in episodes of sepsis but no difference in sepsis or mortality was observed between untreated or treated patients. Among treated patients, no difference in MELD scores was observed between responders or failures and two patients already demonstrating hepatic encephalopathy were rescued from LT by corticosteroids. Six deaths (19%) occurred, all post-transplant. Taken together, the available data suggests (with a very low level of evidence) that all patients should be considered for a trial of corticosteroids at the earliest opportunity and in sufficiently high doses (⩾1 mg/kg) and probably best intravenously [275], but the risk of infections in liver failure has to be kept in mind and may justify the use of prophylactic antibiotics and antifungal agents [206]. LT needs to be considered as an alternative, but the optimal timing is unknown. While no general futility threshold can be identified, it has been shown that failure to improve serum bilirubin, MELD-Na or UKELD within seven days in icteric presentations of AIH has a strong negative prognostic value and should lead to the early consideration of alternative therapeutic strategies including LT [204].

B) Without immediate severityOther patients may experience treatment failure defined by a lack of, or only minimal, improvement in clinical and laboratory features after several weeks of standard treatment but without liver failure. It is a rare event (probably less than 5%) once the original diagnosis has been corroborated and the compliance with therapy confirmed. When lack of compliance or altered metabolism of azathioprine is suspected, measurement of active TGN metabolites may be helpful although the target range has not been fully determined in AIH. Usually 235–450 pmol per 8 × 108 red blood cells by analogy with Crohn’s disease is recommended [[276], [277]]. Recently, TGN concentrations >220 pmol per 8 × 108 red blood cells were shown to be associated with remission in AIH patients [278]. In these non-responding or very slow responding patients, the usual recommendation, based on limited data [207], is to increase predniso(lo)ne to about 60 mg/day (for at least one month) and azathioprine to 2 mg/kg/day if tolerated, as endorsed by the AASLD and BSG guidelines [[34], [40]]. Clinical and laboratory features may improve but most patients remain at risk for drug-related side effects and/or disease progression [207]. In true non-responders, alternative immunosuppression might be required (see below), and expert advice should be sought early for these patients.

Incomplete response

Incomplete response is defined by the occurrence of some improvement in clinical, biochemical and histological parameters but without reaching complete resolution. It includes abnormal liver enzymes or presence of interface hepatitis on a liver biopsy performed in patients with normal liver tests. Once the possibility of non-compliance has again been considered, the optimum strategy remains unclear. In some patients treated with budesonide-based regimen, budesonide (9 mg/day) is insufficient to induce and/or maintain remission and replacement of budesonide with predniso(lo)ne (>20 mg/day initially) should be considered [279]. In other patients treated with predniso(lo)ne-based regimen, increasing the dose of predniso(lo)ne to >10 mg/day is not generally recommended in the long-term because of side effects [40]. Increasing the dose of azathioprine to 2 mg/kg/day, the dose to prevent relapse without corticosteroids [222], together with 5–10 mg/day predniso(lo)ne is a more attractive option. Alternatively, other immunosuppressive drugs may be considered (see below). Whatever regimen is used, a repeat liver biopsy is recommended after a further 18–24 months [40]. The ideal endpoint with complete biochemical and histological resolution may not be attainable in some patients and the goal should be the lowest achievable biochemical activity with a minimum of side effects. While no transaminase threshold has been clearly identified, it is generally assumed that treatments should be adjusted to maintain serum transaminase level below threefold greater than ULN to reduce the likelihood of aggressive interface hepatitis and progression of the disease [[212], [280]]. Histological measures of attenuated disease activity (e.g. HAI <5/18) may be a more reliable guide for these difficult to manage cases.

Alternative drug therapies for unsatisfactory responses

The current choices of second line immunosuppressive therapy include MMF and CNI (cyclosporin or tacrolimus). Many agents have been used with variable success but none have been tested in a randomised controlled trial. Their use in AIH has largely been extrapolated from experience in LT. Their major benefits are potent immunosuppressive activity with a rapid onset of action but these agents exhibit their own side effects: hypertension, renal dysfunction, diabetes mellitus, hyperlipidemia and neurological disturbances for CNIs; diarrhoea, leukopenia and teratogenicity for MMF as well as long-term increased risk of malignancy for both [281]. Unfortunately, available evidence for their use in AIH is mainly based on small, predominantly retrospective case series whose interpretation is hampered by heterogeneity of outcome measures, dosing and indication for therapy (non-response or intolerance).


MMF is an inosine monophosphate inhibitor with both anti-T and -B cell proliferation effects [45]. A prospective but uncontrolled study has examined the role of MMF (1.5–2 g/day) in combination with prednisolone as first line therapy in 59 treatment-naïve AIH patients and found that 88% were responders (normalisation of transaminases and γ-globulins), 37% achieved remission off prednisolone and only two had to discontinue MMF because of severe side effects [45]. Even if these results look promising, further data, including histological outcome, are needed before MMF can be recommended as an alternative first line treatment. A number of case series have reported experience with MMF in patients intolerant to azathioprine or with insufficient response to this drug. Most studies have used 2 g/day in divided doses and, although generally well tolerated, up to one third of patients discontinued MMF due to side effects in some series [282]. In patients intolerant to azathioprine, MMF seems to be an effective alternative with response rates ranging from 43% (12/28) to 88% (8/9) [[283], [284]]. In adult patients with refractory disease, efficacy appears much lower since only 0% (0/12) to 25% (2/8) of patients enter biochemical remission but biochemical improvement occurs in a majority and steroid requirement decreases as well [[283], [284]]. Experience in children is more favourable with a 67% (14/20) response rate [253]. It should be kept in mind that MMF is contraindicated in pregnancy, potentially making other drugs preferable in patients of potential child-bearing age.


CNI have been shown to be effective in a variety of case series in refractory patients. However, it appears that once CNI are started, it is almost always impossible to taper out these drugs again.

a) CyclosporinExperience of cyclosporin in AIH in the paediatric population is mainly as a primary therapy in severe disease or to prevent steroid side effects. Series consistently reported good efficacy with a biochemical response rate ranging from 84 to 100% [[255], [256], [285]]. Evidence for cyclosporin in adults with non-responsive AIH is much smaller but again a high biochemical response rate (⩾80%) was reported at a dose of 2–3 mg/kg/day in small series (with a maximum of six patients) [[286], [287]]. However the number of patients is limited and no long-term reports are available.

b) TacrolimusThe picture is similar with tacrolimus that has been used predominantly as salvage therapy in small series with a maximum of 13 patients or case reports at a dose of 1 to 6 mg/day [[288], [289]]. In the largest single-centre experience with refractory AIH or intolerance to other immunosuppressive drugs, 12/13 patients achieved normalisation of liver tests under tacrolimus (mean trough level of 6 ng/ml) [290]. Furthermore tacrolimus was successful in 7/9 well documented non-responsive severe AIH as reported recently by another centre [204]. Taken together, these data suggest that tacrolimus shows promise in non-responsive AIH and is probably safe although the limitation of all series relates to the short degree of follow-up.

Other immunomodulatory therapy

Other agents have been used with anecdotal evidence of efficacy, including cyclophosphamide (1–1.5 mg/kg/day) [291], methotrexate (7.5 mg/week) [292], rituximab (1000 mg two weeks apart) [293] and infliximab (5 mg/kg at day 0, weeks two and six, and thereafter every four to eight weeks depending on laboratory and clinical course) [294]. Anti-tumour necrosis factor (TNF) antibodies may also induce an immune-mediated liver disease resembling AIH. [[295], [296]]. The use and efficacy of sirolimus has been reported initially in the context of post-transplant AIH [297] and recently for refractory AIH in a non-transplant setting (median through level of 12.5 ng/ml): a sustained >50% fall in ALT was achieved in 4/5 patients including normalisation in two [298]. Main side effects of sirolimus include hyperlipidemia, proteinuria and oedema, but its relatively good safety profile makes it an interesting option. No strong recommendations can be drawn from such small sample sizes and it should be kept in mind that stronger immunosuppression is associated with severe infectious complications, especially in cirrhotic patients [294].

The decision is based on local expertise, AIH severity and patient circumstance. The available evidence does not allow a recommendation as to which of the possible second line drugs should be preferred in an individual patient. Therefore, expert advice should always be sought for applying these experimental second line therapies. In general, at initiation of treatment with non-standard therapy, the doses of current immunosuppression drugs are left unchanged (conversion of azathioprine to MMF excepted) but are gradually decreased thereafter in case of response.


As in any chronic disease, compliance can be a problem during long-term follow-up, especially in paediatric patients entering puberty [299]. Adolescents frequently display poor compliance with medical advice together with poor adherence to therapy and clinic appointments. This may be exacerbated by the cosmetic side effects of steroids. Disease can be even denied in an attempt to be “normal” in a context of need to become self-reliant. As a consequence, non-adherence in this population plays a major role in relapse [300]. Regular monitoring of immunosuppressant drugs is indicated. The management of non-adherence is difficult and relies on a non-judgemental approach. Efforts to improve education, social functioning and behavioural strategies to encourage self-motivation; is better achieved by an active multidisciplinary team including psychologists, youth workers and dedicated nurses who can provide education and support during this difficult period and the transition to adult care. During this time, both paediatric and adult hepatologists should be included. [[301], [302]].

Drug intolerance and side effects

Drug toxicity compels dose reduction or premature discontinuation of the offending drug. Prednisone or prednisolone in AIH has numerous adverse effects (up to 80% after two years) including cosmetic changes (weight gain, facial rounding, and hirsutism), diabetes, emotional instability or psychosis, hypertension and osteoporosis. Severe adverse effects occur mainly at doses >20 mg/day for more than 18 months and lead to treatment discontinuation in about 15% of patients. The combination regimen with azathioprine is associated with much lower occurrence of corticosteroid related adverse events [179]. In the large randomised study by Manns et al. (207 non-cirrhotic patients), comparing prednisone and budesonide (9 mg/day), most of the difference between the two groups at six months was a reduction of steroid side effect in the budesonide group (51.5% vs. 26.0%, respectively) [193]. At the end of the six-month trial, the prednisone group was crossed over to budesonide (6 mg/day) and a 40% reduction of the incidence of steroid side effects was observed in a six-month period. As a result, in predniso(lo)ne responders presenting steroid side effects, a switch to budesonide (6 mg/day) may be considered; alternatively, higher doses of azathioprine (2 mg/kg) should be applied; furthermore conversion of azathioprine to MMF (2 g/day) with subsequent tapering of predniso(lo)ne may be tried, if azathioprine dose is limited by drug toxicity or side effects [[283], [284]].

Up to 25% of patients with AIH develop side effects on azathioprine requiring withdrawal of the drug in about 10% of cases. Side effects are more common in cirrhotic patients. About 5% of patients develop a severe and early reaction with arthralgias, fever, skin rash or pancreatitis within a few days or weeks which warrants its immediate discontinuation. Resolution of symptoms usually occurs in a couple of days. As already discussed above, in patients intolerant to azathioprine, MMF (2 g/day) seems to be a good alternative. 6-MP may be tried even in apparent azathioprine intolerance, as some patients may nonetheless tolerate this active metabolite [303]. Other alternatives are steroid monotherapy in patients with mild disease and little steroid risk factors including good bone density, or those drugs that are also used in case of non-response. The efficacy and tolerance of long-term budesonide monotherapy have not been assessed.

Variant syndromes

The low prevalence of the variant syndromes as well as the lack of universal agreement on definition has made it impractical to perform randomised controlled trials in this setting.

Patients with features of both PBC and AIH

Patients with features of both PBC and AIH seem to have a more severe disease compared to conventional PBC as illustrated by a higher frequency of extensive fibrosis at presentation (despite a younger age in some reports) and most series (but not all) support a worse outcome in terms of biochemical response to UDCA, progression of fibrosis and liver-related mortality [[58], [170], [304], [305]]. Despite the lack of controlled trials, EASL guidelines, based on the results of small series, have recommended adding steroids (predniso(lo)one or budesonide) either at the time of diagnosis of “variant syndrome” or in case of inadequate biochemical response after three-months of UDCA [[58], [60], [304], [306]]. The results of a large multi-centre study (88 patients) have been recently reported: as first line therapy, 30 patients received UDCA alone and 58, a combination of UDCA and immunosuppression (prednisone ± azathioprine); in patients with moderate interface hepatitis, UDCA alone and combination therapy had similar efficacy (80%) in terms of biochemical response whereas in patients with severe hepatitis, efficacy of UDCA alone was much lower (14 vs. 71%, respectively). The presence of extensive fibrosis was associated with a lack of response to the combination therapy but not to UDCA alone; second line immunosuppressive agents (cyclosporine, tacrolimus and MMF) led to biochemical remission in half of the patients who were non-responders to initial immunosuppression [307]. These findings strongly support the use of a combination of UDCA and immunosuppression as first line therapy in PBC patients with severe interface hepatitis. Interestingly, it has been suggested that, in responders, the dose of immunosuppressors in the long-term could be lower and rate of successful withdrawal higher than in classical AIH [[304], [307]]. In UDCA-treated PBC developing AIH (“sequential variant syndrome”) use of immunosuppressive treatment is mandatory [308].

Patients with features of both PSC and AIH

Various results of therapy (usually prednisolone and azathioprine with or without UDCA) have been reported in patients with features of both PSC and AIH [51]. It is difficult to draw any firm conclusions because of the small number of patients, the usually retrospective nature of the studies and the heterogeneity of the regimens. In the paediatric AISC form treated with immunosuppressors, liver biopsies may show improvement in inflammation but cholangiographic appearances may progress and transplant-free survival at ten years (65%) is lower than in AIH (100%) [52]. The combination of UDCA and immunosuppressive therapy may improve liver biochemistry and this approach has been advocated by EASL guidelines [60]. In the series with the most homogeneous regimen (UDCA, prednisolone and azathioprine) including seven patients with a mean follow-up of eight years, a significant fall in transaminases was observed but not in alkaline phosphatase levels, and more interestingly, the Mayo Risk score did not change and transplant-free survival was much better (100%) than that of 34 classical PSC (43%) with the same follow-up and treated with UDCA [309]. However, in the long-term (>10 years), progression towards cirrhosis seems to occur in the majority of patients [64]. Another study reported a similar proportion of patients with AIH-PSC “variant syndrome” (mainly young adults) and AIH achieving AIH biochemical response within one-year of therapy (although the exact combination of azathioprine, prednisone and UDCA varied between individual patients), but the long-term prognosis of patients with an PSC-AIH “variant syndrome” was worse than in AIH without PSC features [310]. Taken together, all these data support the use of UDCA in combination with an immunosuppressive regimen in most patients with features of both PSC and AIH despite the lack of adequate studies [65].

AIH following LT

Recurrent and “de novo” AIH may occur years after grafting and must be distinguished from acute rejection, chronic rejection, viral infection, and drug toxicity. Diagnosis is often challenging because of the lack of a specific marker. Recurrent AIH is reported in about 20–25% of cases [[311], [312]] and is usually managed by increasing the dose of corticosteroids or re-instituting its use with or without azathioprine or MMF [[311], [313]]. In non-responders, conversion of azathioprine/MMF to sirolimus may be successful [297]. Prophylactic use of azathioprine in patients transplanted for AIH has not been evaluated systematically, but appears prudent.

“De novo” AIH has been described in 2–7% of patients undergoing LT for a range of diseases unrelated to autoimmunity, particularly in the paediatric setting [[91], [312]]. The management strategy is similar to that proposed for recurrent AIH [[297], [314]].

Finally, re-transplantation should be considered in patients with recurrent or ‘‘de novo” AIH that is progressing to graft loss, a rare event provided that a prompt therapeutic management has been provided [91].

Treating AIH in the context of liver co-morbidity

Non-alcoholic fatty liver disease (NAFLD)

NAFLD is the hepatic manifestation of the highly prevalent metabolic syndrome which includes obesity and insulin resistance that are known to increase the risk of progression of other liver diseases, especially chronic HCV infection [[315], [316]]. As both AIH and NAFLD may cause persistently elevated aminotransferases and presence of autoantibodies, it is important to have a clear picture of what has to be treated. Liver biopsy is very helpful, and often indispensable, for this distinction. In patients with AIH, the prevalence of the metabolic syndrome or the impact on outcome is unknown, but it is reasonable to presume that the presence of steatohepatitis in patients with AIH will increase the risk of progressive disease. Features of the metabolic syndrome, including diabetes, hypertension and obesity, are exacerbated by corticosteroids. Thus, associated NAFLD should be considered and treated according to current recommendations (lifestyle interventions and pharmacological measures if appropriate). In this population, every effort to use the lowest dose of corticosteroids (combination regimen) should be encouraged.

Chronic viral hepatitis

Although hepatitis B or C should to be excluded before making a diagnosis of AIH, AIH can sometimes develop in patients with HBV or HCV infection either spontaneously or under interferon treatment, and patients with AIH can contract viral infection. Vaccination against hepatitis A virus and HBV should be given to all AIH patients not showing previous vaccination or virus exposure. In patients with hepatitis B or C virus replication and features of AIH at diagnosis, anti-viral (HBV or HCV) interferon-free regimen should be used first and the need for immunosuppressive therapy reassessed after viral eradication or suppression. Reactivation of hepatitis B has been reported during treatment of AIH and baseline HBV serology is recommended for all patients receiving immunosuppressive drugs. According to EASL recommendations [317], HBsAg-positive candidates should be tested for HBV DNA levels and should receive pre-emptive nucleoside/nucleotide analogues administration during immunosuppressive therapy (regardless of HBV DNA levels) and for 12 months after cessation of therapy. In patients with expected lengthy immunosuppression and in all AIH patients, use of either entecavir or tenofovir is recommended.

HIV infection

De novo AIH, as an immune reconstitution, has been described in HIV patients receiving highly active antiretroviral therapy [318]. Liver biopsy findings are critical in establishing AIH diagnosis and discriminating other numerous causes of abnormal liver tests in this setting [319]. Standard immunosuppressive therapy for AIH can be effective but is sometimes associated with life–threatening infections, and treatment for AIH in HIV-infected patients should be individualised after careful consideration of the potential risks and likely benefits [320].