Usefulness of corticosteroids for the treatment of severe and fulminant forms of autoimmune hepatitis


  • Philippe Ichai,

    Corresponding author
    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    • Liver Intensive Care Unit, Centre Hépato-Biliaire, Département des Maladies du Foie, Hôpital Paul Brousse, 94804 Villejuif, France
    Search for more papers by this author
    • Telephone: 0033 (1) 45593000; FAX: 0033 (1) 45593857

  • Jean-Charles Duclos-Vallée,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Catherine Guettier,

    1. Departement de Pathologie, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Sonia Ben Hamida,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Theresa Antonini,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Valérie Delvart,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Faouzi Saliba,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Daniel Azoulay,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Denis Castaing,

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • Didier Samuel

    1. Centre Hépatobiliaire and INSERM Unit 785, Assistance Publique-Hôpitaux de Paris, Université Paris Sud, Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author


Immunosuppressive therapy, and particularly corticosteroids with or without azathioprine, can achieve a remission in more than 80% of patients with autoimmune hepatitis (AIH). By contrast, the usefulness of corticosteroid therapy in severe forms of AIH remains a subject of debate. Between 1986 and 2005, 16 patients (14 females, 2 males; mean age: 36.6 ± 13.1 yr) presenting with acute, severe, or fulminant disease due to type 1 AIH (n = 13) or type 2 AIH (n = 3) were admitted to our liver intensive care unit. At admission, 10 of 16 (62.5%) patients presented with encephalopathy. Median international normalized ratio (INR), bilirubin, alanine aminotransferase (ALT), and creatinine values were 5.36 (range, 1.7-12.2), 425 μmol/L (range, 278-850), 678 IU/L (range, 60-2867), and 72 μmol/L (range, 52-133), respectively. A total of 12 patients received corticosteroid therapy: 8 had started in the referring center a median of 2.5 days (range, 1-89) previously, and this therapy was initiated in 4 patients at their admission to our unit (median: 2 days; range: 0-5). Four patients were not treated because of a rapid deterioration in their AIH. Before treatment, 4 of 12 patients had been suffering from encephalopathy. The median duration of corticosteroid therapy was 7 days (range: 2-135). Of 16 patients, 13 underwent liver transplantation (LT) (81%), at which time all were encephalopathic. Median values for INR, total bilirubin, and ALT were 7.2 (range: 3.3-15.9), 400 μmol/L (range: 301-550), and 706 IU/L (range: 69-1,932), respectively, at the time of transplantation. All patients treated with corticosteroids had experienced a clinical (encephalopathy) and biochemical (Model for End-Stage Liver Disease [MELD] score) deterioration at the time of transplantation. Histological findings did not reveal any features of underlying chronic liver disease. Of the 13 patients undergoing transplantation, 10 had received prior corticosteroid therapy. Of the 2 nontransplanted patients treated with corticosteroids, a clinical improvement was observed in only 1 patient. Severe septic complications occurred in 3 patients under corticosteroid therapy (gram-negative septicemia n = 2; disseminated aspergillus n = 1). Nine of the treated patients are still alive; 1 died after liver transplantation (LT) (recurrence of AIH, acute pancreatitis, sepsis), 1 survived without LT, and 1 died without LT. Among the untreated patients, 3 survived after LT and 1 died without LT. In conclusion, corticosteroid therapy is of little benefit in severe and fulminant forms of AIH; it may favor septic complications and should not delay LT. Liver Transpl, 2007. © 2007 AASLD.

Autoimmune hepatitis (AIH) is a liver disease of unknown etiology that often affects young females. It is a form of chronic hepatitis characterized by hypergammaglobulinemia and serum antibodies, and is unrelated to viral infection, hepatotoxic drugs, or hereditary disorders. The natural outcome is generally poor, with cirrhosis at presentation in 50 to 90% of patients. In chronic forms, preferred treatment consists in corticosteroids and azathioprine.1 Remission is achieved in over 70% of cases and long-term treatment with azathioprine, with or without prednisolone, can prevent relapse.2, 3 However, the usefulness of immunosuppressive therapy in severe and fulminant forms of AIH has not been demonstrated. Moreover, the fact that corticosteroid therapy might obviate the need for liver transplantation (LT) has not been proven. We decided to perform a retrospective analysis of our cohort of patients with severe forms of AIH. The endpoint of this study was to determinate the effectiveness of using steroid therapy in severe and fulminant forms of AIH.


AIH, autoimmune hepatitis; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; INR, international normalized ratio; ALT, alanine aminotransferase.


A total of 16 patients (14 female, 2 male) with a mean age of 36.6 ± 13.1 yr (range: 16-67 yr) were referred to our liver intensive care unit for severe forms of AIH between 1986 and 2005. The diagnosis of AIH was based on the presence of at least 3 of the following diagnostic criteria (International Autoimmune Hepatitis Group)4: 1) autoantibodies at significant titers; 2) gamma globulin levels above 20 gm/L; or 3) liver biopsy revealing features compatible with chronic hepatitis. Type 1 AIH was defined by the presence of antinuclear antibodies and/or antismooth muscle antibodies, and type II AIH by the presence of type 1 anti-liver-kidney microsome antibodies and/or antiliver cytosol antibodies. Autoantibodies were detected by indirect immunofluorescence using cryostat tissue fractions according to standard procedures.5 A total of 13 patients were diagnosed as suffering from type 1 AIH and 3 from type 2 AIH. No autoantibodies were detected in four patients, but the diagnosis of AIH was based on high immunoglobulin G levels and the presence of a marked lymphoplasmacytic infiltrate on histological examination of a specimen obtained by transjugular liver biopsy. Liver histology findings were assessed on transjugular liver biopsy specimens in 7 patients at admission. All native livers underwent histological reappraisal, and histological features compatible with AIH, such as the presence of lymphoplasmacytic infiltrate and interface hepatitis, were described. Histological examinations of native liver were performed a median of 6 days (range: 2-58) after the initiation of corticosteroid therapy (Table 1). No patients had been consuming excessive quantities of alcohol or drugs before they were admitted. A diagnosis of Wilson's disease was excluded in all patients by the verification of serum copper, urinary copper, and serum ceruloplasmin levels. In patients enrolled before 1998 (patients 1, 6, 8, 10, and 14), INR was not available. This parameter (WR) was calculated as follows: [prothrombin time patient (seconds) / prothrombin time control (seconds)]ISI, where ISI is the international sensitivity index. The international sensitivity index to thromboplastin was estimated with a very strong accuracy at 1.8 by our laboratory. In these patients, Model for End-Stage Liver Disease (MELD) score was determined according the calculation of INR, as described before.

Table 1. Clinical, Biological and Histological Characteristics at Admission of the 16 Patients With Severe Autoimmune Hepatitis
PatientsClinical and biological parametersImmunologicalHistological (biopsy admission1, native liver2, post mortem biopsy3
Age (yr)/gender (M/F)Grade HETotal bilirubin (μmol/L)ALT (IU/L)INRFact V (%)MELD score (b)IgG (g/L)AutoAb titerNecrosis intensityInflammatory infiltrateType of cells
  • NOTE: No fibrosis was observed in any patient from histological examinations performed on biopsies at admission, in native liver and/or from post mortem biopsy.

  • Abbreviations: HE, Hepatic encephalopathy; P, plasmocytes; L, lymphocyte; PN, polynuclear neutrophils; AutoAb titre, autoantibody titre; ALT, Alanine aminotransferase; INR, International Normalized Ratio; M, male; F, female; SMA, smooth muscle antibodies; ANA, antinuclear antibodies; HE, hepatic encephalopathy; Fact V, Factor V; IgG, immunoglobin G.

  • Sub-massive necrosis (50–70%).

  • Massive necrosis (>80%).

  • §


  • Severe.

  • //


136/F22786058.1134033-2§2L//, PN§
238/M04004691.7662433.2SMA: 1/801§1L§, P//, PN//
348/F02892,8673302917ANA: 1/6401, 21.2L1, 2, P1, 2, PN§1, 2
         SMA: 1/160   
468/F34472183.3333423SMA: 1/801//3L//
537/F04441,9266333924-1//1, 2L//1, 2, PN§1
          2 PN//2
636/F33888307.274016ANA: 1/502§2L§
         SMA: 1/50   
723/F2850923.02313419LKMI:1/6403//3L//, P//, PN//
844/F44202007.2154120SMA: 1/1601, 2§2L§1, 2
921/F04911,8065.4303719SMA: 1/3202//2L//, P//
1144/F14307505.3303722SMA: 1/6401, 2//1, 2L§1, 2, PN//1, 2, P//1, 2
1321/F04508213.02523113.5LKMI:1/6401, 2//1, 02L§1
1443/F43435934203322.5ANA:1/1002§2L§ PN§
1534/M32939396233739ANA:1/16022L§, P§
         SMA: 1/160   
1621/F05091,10112.2114739.2ANA:1/8022L§, P§

Before corticosteroid therapy was initiated, an absence of sepsis was confirmed by negative cultures of blood samples, ascites fluids, and urine specimens, and by chest X-ray.


Fulminant and subfulminant hepatitis were defined as acute hepatitis complicated by acute liver failure and hepatic encephalopathy occurring less than 2 weeks or between 2 weeks and 12 weeks, respectively, after the onset of jaundice.6 Encephalopathy was graded from 1 (least severe) to 4 (most severe).7 Coma was graded from 0 to 3 and started in encephalopathy stage 4. Indications for LT were taken as the presence of stage 3 or 4 hepatic encephalopathy associated with either plasma protein factor V levels lower than 20% of normal in patients below the age of 30 yr, or factor V levels lower than 30% of normal in patients over the age of 30 yr. All patients were managed according to a standardized protocol.8–10 Artificial liver support with albumin dialysis was ensured in 1 patient who presented a contraindication to LT.11, 12 A total of 6 LT patients received primary immunosuppressive therapy with prednisone, cyclosporine, and azathioprine, whereas the other 7 were given prednisone, tacrolimus and among them, mycophenolate mofetil was added in 2. In addition, 2 patients were treated with antilymphocyte serum and 1 with Basiliximab as induction therapy. Prophylaxis for Pneumocystis carinii infection and cytomegalovirus infection was ensured by trimethoprime sulfate (3 patients) and ganciclovir (4 patients). Bacterial infection was prevented by the use of piperacillin in all patients during the perioperative period (2 days). In 3 patients (patients 5, 11, and 15), fungal infections were prevented by using of amphotericin B lipid complex (Abelcet, 1 mg/kg/day for 7 days, then 2.5 mg/kg twice per week for 3 weeks) after LT. Curative treatment by adapted antibiotic was performed after occurrence of sepsis.


Statistical analysis with the StatView statistical software (version 5.0, StatView, Cary, NC) was performed using the Wilcoxon test. The various liver test parameters (total bilirubin, ALT, INR) and prognostic scores (MELD) were compared at the onset of disease, at admission, prior to corticosteroid therapy and at the time of LT.


Characteristics at Admission

At admission, 10 of 16 (62.5%) patients suffered from encephalopathy: stage 1, n = 1; stage 2, n = 2; stage 3, n = 4; and stage 4, n = 3 (coma grade 1: 2 patients; grade 3: 1 patient). Median values for total INR, bilirubin, and ALT (normal < 45 IU/L) were 5.36 (range: 1.7-12.2), 425 μmol/L (range: 278-850), and 678 IU/L (range: 60-2,867), respectively. The mean creatinine level was 81 ± 23 μmol/L. The clinical, biological, and histological characteristics of the 16 patients are summarized in Table 1. The median delay between the onset of jaundice and admission to the liver intensive care unit was 21 days (range: 6-148 days). At admission, 1 patient (patient 4) had been receiving mechanical ventilation support for 1 day.

Of the 16 patients studied, 8 were receiving corticosteroid therapy at admission. Prednisone (1 mg/kg body weight, intravenously) had been initiated in these 8 patients a median of 2.5 days previously (range: 1-89). One of these patients had received corticosteroids and azathioprine (1 mg/kg/day), but this had been discontinued after 8 days because of severe hematological intolerance. The patient was then switched to cyclosporine (4 mg/kg/day), which was maintained until admission into our unit, when a diagnosis of disseminated Aspergillus fumigatus was made. In 4 other patients, prednisolone (1 mg/kg/ body weight) was started 2 days (range: 0-5) after admission. Liver biopsy findings were available for 6 patients. In 5 patients, these findings had been determined prior to corticosteroid therapy, the median interval elapsing being 1 day (range, 0-20; patients 2, 3, 8, 11, and 13). In 1 patient, the liver biopsy was performed 36 days after the initiation of corticosteroid therapy. Necrosis was severe in 1 patient and moderate in 5 patients. The median weight of native liver available in transplanted patients was 620 gm (range: 440-1,081). Details of the histological characteristics of native liver are shown Table 1. None of the 16 patients displayed morphologic evidence of underlying chronic liver disease. Necrosis was severe in 62.5% of patients and moderate in 37.5% of patients. Plasma cell infiltration was observed in 7 patients (43.8%).


Of the 16 patients, 13 (81%) underwent LT within a median of 2 days (range: 0-60) of admission (Fig. 1, Table 2). The decision concerning LT was made because of the onset of encephalopathy in 5 patients (grade 3 in 3 patients, and grade 4 in 2 patients), or an increase in and/or persistence of encephalopathy in the remaining 8 patients with elevated INR values. Three patients were receiving mechanical ventilation prior to LT (patients 4, 5, and 15). At the time of LT, median total bilirubin, INR, factor V, and ALT values were 403 μmol/L (range: 301-940), 8.5 (range: 3.3-15.9), 20% (range: 9-33), and 605.5 IU/L (range: 68-1932), respectively. The median creatinine level was 72 μmol/L (range: 55-103) at the time of LT.

Figure 1.

Outcome in our cohort of 16 patients referred to the intensive care unit because of severe AIH.

Table 2. Outcome of Patients Treated or Not Treated With Corticosteroids
PatientsSteroid therapy (Y/N)Duration of treatment prior adm (days)Duration of steroid therapy before LT (days)Delay HE -therapy (days)LTHE gradeComa stageBilirubin (μmol/L)INRFactor V (%)ALT (IU/L)Alive (Y/N)
  • Abbreviations: adm, admission; HE, hepatic encephalopathy; INR, International Normalized Ratio; Y, yes; N, no; NA, not applicable.

  • *

    Alive without LT.


Outcome of Patients Without Corticosteroid Therapy (n = 4)

Of the 16 patients, 4 were not treated with corticosteroids, either because of a rapid deterioration in their clinical status (patient 4), or because AIH had not been diagnosed at the time of LT (patients 9 and 12), or because they underwent surgery before drug therapy could be instituted (patient 16). Three of these patients underwent transplantation 24 hours (2 patients) and 48 hours (1 patient) after admission. LT was not performed in the remaining patient (patient 4), because of severe sepsis and multiple organ failure.

Outcome of Patients Receiving Corticosteroid Therapy (n = 12)

The median duration of corticosteroid therapy was 7 days (range: 2-135), with a mean dosage of 1 mg/kg/day. Encephalopathy occurred in 7 patients treated with corticosteroids, the median time to onset being 18 days (range: 1-86) after initiation. All were suffering from encephalopathy at the time of LT. A total of 10 patients underwent LT. The median interval between admission and LT was 3 days (range: 1-60). Between the time of starting therapy and the time of LT, changes were seen in the encephalopathy stage from 0 to 3 (P = 0.006), bilirubin levels from 382 to 377.5 μmol/L (P = not significant), ALT values from 960 to 505 IU/L, INR values from 3.9 to 6.9 (P = 0.02), and the MELD score from 32.5 to 40 (P = 0.01). Overall, between the initiation of therapy and LT, the clinical status and liver function of patients deteriorated significantly (P = 0.01).

Two of the 12 patients treated did not undergo transplantation. The clinical condition and liver function parameters of one patient improved 3 days after initiating corticosteroid therapy (patient 2). At admission he was jaundiced, with a bilirubin level of 400 μmol/L, a prothrombin time of 49%, and a factor V value of 66%, and ALT values of 469 IU/L (normal < 43 IU/L). He received corticosteroids at a dose of 1 mg/kg/day. After 12 months of follow-up, he is now well and has not experienced any recurrence of autoimmune disease; corticosteroid therapy has gradually been tapered. The second patient (patient 7) was suffering from disseminated (pulmonary and ocular) aspergillosis at admission into our unit. Her clinical condition gradually deteriorated and she died 26 days later, despite the withdrawal of corticosteroid therapy. Infections occurred in 5 patients during corticosteroid therapy (patients 2, 7, 10, 13, and 15), and included: spontaneous bacterial peritonitis (Escherichia coli) (patient 15); urinary tract infection (Proteus mirabilis) (patient 13), endometritis (Klebsiella pneumonae) (patient 10), a positive blood culture for methicillin-resistant Staphylococcus aureus (patient 2), and disseminated aspergillosis (patient 7). One patient died during the perioperative period from multiple organ failure secondary to sepsis (patient 5).

Three courses of extracorporeal liver support (Molecular Adsorbent Recycling System, Gambro® AB, Lund, Sweden) were administered in 1 patient, 2 days after admission (patient 7). Bilirubin levels decreased to reach a mean level of 196 μmol/L, but no elevation of the prothrombin time was noted after each session.

Posttransplantation Outcome

A total of 12 patients are still alive with a mean posttransplantation follow-up period of 9 yr (range: 0.6-18.6 yr). Of these, 9 had received corticosteroid therapy prior to LT. One patient died 9.6 months after LT (patient 5) because of a severe recurrence of autoimmune disease on the liver graft, acute pancreatitis, and severe sepsis.


AIH was the first chronic liver disease in which drug therapy was shown to prolong survival. The standard initial therapy used to induce remission is either prednisone monotherapy or combination therapy with prednisone and azathioprine. Using this conventional treatment, remission (defined as a complete biochemical and histological response) can be achieved in most patients within a few weeks. It is now generally accepted that patients with mild disease have a better response.3 As for severe forms of AIH, and particularly inaugural severe forms, the management of drug therapy is controversial. It is not clear whether the introduction of corticosteroid therapy can induce a benefit and thus obviate the need for LT. Moreover, the decision to initiate immunosuppressant drugs must be counterbalanced by the risk of septic complications.13

This retrospective series concerned 16 patients with an acute and severe inaugural presentation of AIH, who were admitted to our unit between 1986 and 2005. These forms account for a small proportion of acute liver failure patients (3.7% of all such patients admitted to our unit for this reason over the same period). The clinical outcome in this series involved a clinical and biochemical deterioration in most patients. Of the 16 patients studied, 13 underwent LT. Their deterioration was rapid and we observed a significant increase in the MELD score during the first days after admission, although use of this score has not been validated in acute liver failure. We should also emphasize that these patients displayed very poor liver function parameters at admission, with a median MELD score of 37 (range: 24-47) and marked cholestasis, giving rise to median bilirubin levels of 425 μmol/L (range: 278-850). Clearly, the initiation of corticosteroid therapy did not prevent LT in most of these patients. In patients who received corticosteroid therapy, the mean dosage was 1.3 mg/kg/day at the time of treatment initiation. Most of them had received corticosteroids very soon after the onset of early symptoms: immunosuppressant therapy was administered in 8 of 12 patients within a few hours of admission. The duration of therapy was longer than 3 months among the 12 patients treated. Only 1 patient (patient 2) treated with corticosteroids did not receive a transplant because of improvement. It should however be pointed out that this patient never presented with encephalopathy during the course of his disease, and his peak INR value was 1.7, suggestive of a less severe condition.

In different patient series, the response to corticosteroid therapy in a setting of acute AIH has fluctuated between 36% and 100% of the patients treated (Table 3).14, 16–22, 26, 27 Of those receiving corticosteroid therapy, 19% underwent LT, 11% died without LT, 48% improved, and 2% were listed for LT. However, these studies did not provide any information concerning the severity of their condition, and it is difficult to conclude whether corticosteroid therapy was of value in severe forms. Moreover, no details were given concerning dosage, the timing of treatment initiation, and the delay before it was introduced. Subfulminant or fulminant hepatic failure may constitute the first manifestation of disease in both types of AIH (types 1 and 2). However, in the principal patient series, type 1 AIH was observed more frequently than type 2 AIH.3 In our study, 13 of the 16 patients presented with type 1 fulminant AIH.

Table 3. Description of the Patient Series Concerning the Management of Severe and Acute Autoimmune Hepatitis
AuthorsNMean age (yr)Bilirubin adm (μmol/L)ALT adm (IU/L)PT (INR)HEIMS (n)IMS therapyImprovement under steroid (delay)LTOutcome in pt with steroid therapy (A, alive D, died)
  • Abbreviations: adm, admission; HE, Hepatic encephalopathy; IMS, immunosuppressant drugs; P, prednisone; A, azathioprine; CyA, cyclosporine; CR, complete response; R, relapse; NR, no response; TP, temporary response; N, normal; NA, not applicable; n, number.

  • *

    Temporary improvement, then deterioration 7 weeks later.

  • Patient listed for LT.

  • 17 of 22 patients without cirrhosis.

  • §

    7 of 8 patients with grade 2 hepatic encephalopathy: 1 patient with grade 1.

  • Azathioprine, tacrolimus, mycophenolate mofetil, 6 mercaptopurine, cyclosporine.

  • 6/28 patients with grade 1 encephalopathy, 12/28 patients grade 2–3 and 10/28 grade 4.

Amontree et al. (1989)166462921,043NA1/66P: 26/6 (1 week)0A : 6
        P + A: 4   
Maggiore et al. (1990)173223827 N21.7%03P: 13/2/NA0A:3
        P + A: 2   
Bosh et al. (1991)2119243 20.5% 1P + A1*1LT:1
Nikias et al. (1994)221244106 NANA12P:2/12CR: 9/12, (R:7/9)0A:11
        P + A:10/12NR: 2/12 D:1
Herzog et al. (1997)19315251901(4.6)0/0/3–2P + A:103LT:2
      3/2/3 P+ CyA:1   
Viruet and Torres (1998)18136410 (3,8)01P1/1 (10 days)0A:1
Abe et al. (2001)151344160600.3NANA11P: 9CR: 8/110A : 8
        P + A:2NR:3/11 D:2
           Listed for LT:1
Okano et al. (2003)269502361,040NANA9P: 99/91NA
Ferrari et al. (2004)27224013525.3 NNANANANANANANA
Kessler et al. (2004)1410402901,17910-15%8§10P: 10CR: 4/103A : 4
        Others: 9NR: 6/10 D:2
           Listed for LT : 1
Villamil et al. (2004)202841398 30%2825P in allCR: 9/2512A:9

Our patient series also enabled the collection of histological data. All patients underwent a liver biopsy, which revealed a submassive severity of necrosis in 6 patients (37%), and massive necrosis in 10 patients (63%). No fibrosis was present, so we can conclude that these patients were free of underlying chronic liver disease. This disease represents a particular form of severe and rapid evolution of AIH.23 Minimal to severe inflammation was observed in all patients. Interestingly, patients with a severe inflammatory infiltrate (n = 3) did not benefit from corticosteroid therapy. However, it was difficult to reach any conclusions based on histological findings because the techniques employed differed (transjugular liver biopsy, native liver, and postmortem biopsy), the biopsies were collected at different times and not all patients were treated with corticosteroids.

In a recent series of 31 patients with fulminant forms of AIH, reported by Villamil et al.,20 25 of whom received corticosteroid therapy, pejorative prognostic factors under univariate analysis were a prothrombin level <20%, encephalopathy at admission, histologically-proven massive or submassive necrosis, type 2 AIH, and an absence of improvement in liver function with a 20% increase in the prothrombin time on the third day after the initiation of corticosteroid therapy. In our series, all patients displayed at least 2 of these criteria: the mean prothrombin level was 22%, none of the patients experienced an improvement in liver function with a 10% increase in the prothrombin time, and 10 of the 16 patients had encephalopathy at admission. The results of our study argue against the hypothesis that corticosteroid therapy could avoid LT in severe and fulminant hepatitis in a setting of autoimmune liver disease. Moreover, severe complications such as infections may occur.24, 25 It is not clear whether sepsis was directly related to corticosteroid therapy or to liver failure. Among 5 patients (patients 2, 7, 10, 13, and 15) who developed sepsis, 2 of these (patient 7 and 10, respectively) received steroid therapy or other immunosuppressive agent for a long time prior the admission in our Liver Intensive Care Unit (89 and 33 days, respectively). In our series, 1 patient died from severe sepsis as a result of gram negative septicemia, 10 months after LT. Indeed, in a subset of patients who have been treated for a long time by steroid therapy, the introduction of antibiotherapy and antifungal therapy must be discussed. In the future, strict severity criteria need to be defined for these severe forms of liver disease in order to produce guidelines for the management of these patients: LT or corticosteroid therapy. Moreover, it is necessary to develop a dynamic score to calculate potential improvements in liver function in the context of corticosteroid therapy.

Our conclusion is that patients presenting with a very severe or fulminant form of AIH should be referred to a liver intensive care unit, because LT, which works well in this subset of patients, must be proposed as soon as possible, and corticosteroid therapy is ineffective in most cases.