Immunoproteomic Analysis of Potentially Severe Non-Graft-Versus-Host Disease Hepatitis After Allogenic Bone Marrow Transplantation


  • Elvire Beleoken,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    Search for more papers by this author
  • Rodolphe Sobesky,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Centre Hépato-Biliaire, Villejuif, France
    Search for more papers by this author
  • Jean-Pierre Le Caer,

    1. Centre de Recherche de Gif, Institut de Chimie des Substances Naturelles, CNRS, Gif-sur-Yvette, France
    Search for more papers by this author
  • François Le Naour,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    Search for more papers by this author
  • Mylène Sebagh,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Laboratoire Anatomie Pathologie, Villejuif, France
    Search for more papers by this author
  • Nicolas Moniaux,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    Search for more papers by this author
  • Bruno Roche,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Centre Hépato-Biliaire, Villejuif, France
    Search for more papers by this author
  • Mohammad Zahid Mustafa,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    Search for more papers by this author
  • Catherine Guettier,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Laboratoire Anatomie Pathologie, Villejuif, France
    Search for more papers by this author
  • Catherine Johanet,

    1. INSERM, Unité 785, Villejuif, France
    2. AP-HP, Hôpital Saint-Antoine, Laboratoire Immunologie, Paris, France
    3. UPMC, Paris, France
    Search for more papers by this author
  • Didier Samuel,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Centre Hépato-Biliaire, Villejuif, France
    Search for more papers by this author
  • Jean-Henri Bouhris,

    1. Institut Gustave Roussy, Unité de Greffe de Moelle, Villejuif, France
    Search for more papers by this author
  • Jean-Charles Duclos-Vallee,

    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP Hôpital Paul Brousse, Centre Hépato-Biliaire, Villejuif, France
    Search for more papers by this author
  • Eric Ballot

    Corresponding author
    1. INSERM, Unité 785, Villejuif, France
    2. Université Paris-Sud, UMR-S 785, Villejuif, France
    3. AP-HP, Hôpital Saint-Antoine, Laboratoire Immunologie, Paris, France
    • INSERM U 785, Hôpital Paul Brousse, Centre Hépato-Biliaire, 14 Avenue Paul Vaillant-Couturier, 94807 Villejuif, France;
    Search for more papers by this author

  • Potential conflict of interest: Dr. Samuel consults for Novartis and DSG.


The development of potentially severe non-graft-versus-host disease (GVHD) hepatitis resembling autoimmune hepatitis (AIH) has been reported after bone marrow transplantation (BMT). The aim of this study was to better characterize this form of hepatitis, particularly through the identification of autoantigens recognized by patient sera. Five patients who received an allogeneic BMT for the treatment of hematological diseases developed liver dysfunction with histological features suggestive of AIH. Before and during the onset of hepatic dysfunction, sera were tested on immunoblottings performed with cytosolic, microsomal, mitochondrial, and nuclear proteins from rat liver homogenate and resolved by two-dimensional electrophoresis. Antigenic targets were identified by mass spectrometry. During the year that followed BMT, all patients presented with GVHD. Acute hepatitis then occurred after the withdrawal, or during the tapering, of immunosuppressive therapy. At that time, no patients had a history of liver toxic drug absorption, patent viral infection, or any histopathological findings consistent with GVHD. Immunoreactive spots stained by sera collected at the time of hepatic dysfunction were more numerous and more intensely expressed than those stained by sera collected before. Considerable patient-dependent pattern heterogeneity was observed. Among the 259 spots stained exclusively by sera collected at the time of hepatitis, a total of 240 spots were identified, corresponding to 103 different proteins. Twelve of them were recognized by sera from 3 patients. Conclusions: This is the first immunological description of potentially severe non-GVHD hepatitis occurring after BMT, determined using a proteomic approach and enabling a discussion of the mechanisms that transform an alloimmune reaction into an autoimmune response. Any decision to withdraw immunosuppression after allogeneic BMT should be made with caution. (HEPATOLOGY 2013)


2D, two-dimensional; Abs, antibodies; AEBSF, 4-(2-aminoethyl) benzenesulfonyl fluoride; AIH, autoimmune hepatitis; AMA, antimitochondrial antibodies; ANA, antinuclear antibodies; AST, aminoaspartate transferase; ATP, adenosine triphosphate; BMT, bone marrow transplantationp; CAT, catalase; CHCA, cyano-4-hydroxycinnamic acid; CMV, cytomegalovirus; CoA, coenzyme A; DTT, dithiothreitol EBV, Epstein Barr virus; GVHD, graft-versus-host disease; HAV, hepatitis A virus; HBV, hepatitis B virus; HCV, hepatitis C virus; HEV, hepatitis E virus; HHV, human herpes virus; HLA, human leukocyte antigen; HPLC, high-performance liquid chromatography; HSP, heat shock protein; HSV, herpes simplex virus; Ig, immunoglobulin; IIF, indirect immunofluorescence; IPG, immobilized pH gradient; IS, immunosuppression; GGT, gamma-glutamyl transferase; LC1, liver cytosol type 1; LKM-1, liver-kidney microsome type 1; MS, mass spectrometry; MALDI, matrix-assisted laser desorption-ionization; MMF, mycophenolate mofetil; PCR, polymerase chain reaction; SMA, anti–smooth muscle antibody; TFA, trifluoroacetic acid; TOF, time of flight; TLRs, Toll-like receptors.

Allogeneic bone marrow transplantation (BMT) is a procedure used to treat severe hematological, immunological, and inherited metabolic disorders, such as leukemia, lymphoma, autoimmune diseases, or primary immunodeficiencies.1 Initially restricted to patients with human leukocyte antigen (HLA)-identical sibling donors, BMT is currently performed using marrow from unrelated or HLA-mismatched related donors. Although BMT is a life-saving procedure, the use of nonidentical HLA donors favors the development of serious, life-threatening complications. Although cytopenia, thyroid diseases, and myasthenia gravis are autoimmune phenomena that can develop after BMT,2 the most common complications are graft rejection by recipient cells and graft-versus-host disease (GVHD), caused by donor T cells attacking recipient tissue. GVHD staging and patient survival are largely dependent on whether its complications involve the skin, liver, lung, or intestine.

Liver complications after BMT have multiple origins, such as viral acute hepatitis (i.e., cytomegalovirus [CMV] infection), drug consumption, iron overload, veno-occlusive disease, nodular regenerative hyperplasia, and, in the vast majority of cases, acute or chronic GVHD.3,4 The incidence and severity of liver GVHD vary as a function of the age or gender of the patient and the degree of HLA mismatch. Although most patients survive the disease without long-term disabling side effects, liver GVHD can be fatal. Patients presenting with two or more different liver diseases are not rare.

Despite advances in the management of patients undergoing BMT, the risk of developing liver GVHD post-BMT after the withdrawal of immunosuppressive treatment remains a current issue. Some studies in the literature have reported cases of BMT followed by non-GVHD liver dysfunction with the occurrence of autoantobodies.5-10 Advances in proteomic analysis currently provide an opportunity to better characterize and understand the pathogenesis of autoimmune diseases, including those that affect the liver,11-13 and to identify markers for early diagnosis and follow-up.

The aim of this study was therefore to report on some cases of potentially severe non-GVHD hepatitis and to characterize the antigenic targets recognized by antibodies detected in the sera of these patients using serological proteome analysis. These severe forms of non-GVHD hepatitis are poorly described in the literature and a clearer understanding of them may enable adaptations to the management of immunosuppression (IS) after BMT.

Patients and Methods

Selected Patients.

Of the 235 patients who underwent an allogeneic BMT in a bone marrow transplant center (Institut Gustave Roussy, Villejuif, France) between 2004 and 2009, 5 (2.1%) developed hepatic dysfunctions that mimicked autoimmune hepatitis (AIH). This group of patients included 1 woman and 4 men, with a mean age of 48.2 years (range, 43-51). The detailed clinical characteristics of the transplanted patients are presented in Table 1. The donor/recipient genders differed in 1 case (male recipient/female donor). In patient P1, HLA A, B, DR, and DQ were compatible, and there was one DP mismatch (the HLA recipient/donor status was A 0201 0301/0201 0301, B 0702 2705/0702 2705, C 0102 0702/0102 0702, DRB1 0801 1101/0801 1101, DQB1 0402 0301/0402 0301, and DPB1 021 0401/0201 0402). In patient P2, there was no HLA mismatch. The HLA recipient/donor status was A 3 33/3 33, B 7 71/7 71, DRB1 0815/0815, and DQB1 0506/0506. In patient P3, there was no HLA mismatch, and the recipient/donor status was A 02/03, B 07/51, C 07/14, DRB1 0815/0815, and DQB1 04/06. There was no HLA mismatch in patient P4, and the recipient/donor status was A 29/29, B 44/44, DRB1 01 07/0101 0701, and DQB1 02 05/02 05. In patient P5, there was no HLA mismatch, and the recipient/donor status was A 3, B 14, 35*01, *13, and *05.

Table 1. Biological and Pathological Features of Patients With Non-GVHD Hepatitis
  1. Abbreviations: ALT alanine aminotransferase; CS, cyclosporine; PT, prothrombin time; ++, moderate; +, mild.

Sex/age at the BMTF/51M/52M/53M/42M/43
PathologyMyelodysplastic syndromeChronic lymphocytic leukemiaChronic lymphocytic leukemiaAcute myeloblastic leukemia 2Hodgkin's lymphoma
HLA mismatchOne DPNoneNoneNoneNone
GVHD onset after BMT10 days (hands, skin, face)3 months (rash on torso and hands)10 months (rash on torso)9 months (digestive)2 months (suspicion cutaneous GVHD) plus 12 months (cutaneous GVHD)
GVHD therapyCS+corticosteroidMMF+CS+corticosteroidCS+corticosteroidCS+corticosteroidCS+corticosteroid
Onset of hepatic dysfunction after withdrawal of IS1 month after IS withdrawal1 month after IS withdrawal3 months after IS withdrawalDuring withdrawalOne week after IS withdrawal hepatic GVHD plus dual hepatitis plus hemorrhagic necrosis
Onset of hepatic dysfunction after BMT13 months12 months13 months12 months6 months
Biochemical features     
 PT 37%–100% AST 13-72× ALT up to 20×. Bilirubin 82-270 µmol/L, GGT 455-1,968 IU/L
IgG 24.5 g/LIgG 6.7 g/LIgG 3.38 g/LIgG 6 g/LIgG 24.4 g/L
Viral markersHCV positive,HCV negative
before BMTHCV RNA negative 
After BMTHAV, HBV, CMV, HSV, HHV6, EBV negative
ANA (IIF)Negative before and at the onset of hepatic disorders1/80 at the onset of hepatic disorders1/640 at the onset of hepatic disordersNegative before and at the onset of hepatic disorders
Other hepatic markers (IIF)Anti-SMA, anti-LKM1, antimitochondrial negative before and at the onset of hepatic disorders
Histological features     
 Activity localizationInterface, centrolobularLobularInterfaceInterface, lobularInterface, lobular
 Veno-occlusive disease00000
 GVHD00Mild cholangitis00

After BMT, all the selected patients received standard therapy to prevent GVHD (i.e., cyclosporine and corticosteroids), sometimes combined with another immunosuppressive therapy, such as mycophenolate mofetil (MMF). All patients developed GVHD between 10 days and 12 months after BMT (median delay: approximately 7 months). Cutaneous signs were detected in 4 patients and digestive disorders in 1.

From 6 to 13 months after BMT (average, 11.2), all 5 patients developed acute hepatitis during the withdrawal (patients P1, P2, P3, and P5) or tapering (patient P4) of immunosuppressive therapy. The histological, biological, and immunological features of these patients are described below.

Two control groups for this study were composed of sera from 3 patients with acetaminophen hepatitis and 3 with well-characterized AIH. Their clinical and biological features are summarized in Supporting Table 1.

Pathological Examination.

Liver tissue specimens were obtained from percutaneous or transjugular liver biopsy at the onset of hepatic dysfunction. The biopsy samples were embedded in paraffin for routine staining techniques, including hematoxylin and eosin, Masson's trichrome, and picrosirius red for collagen. Fibrosis and inflammatory activity (including the amount of periportal piecemeal necrosis, lobular necrosis, and portal inflammation) were evaluated separately. In addition, the most characteristic histological features of chronic hepatitis and AIH were recorded, including plasma cell infiltrates (semiquantitatively evaluated as +++ severe, ++ moderate, or + mild), lymphoid follicles, rosette formation, acidophilic degeneration, parenchymal collapse, hepatocellular ballooning, multinucleated hepatocytes, intrasinusoidal infiltrates of lymphocytes, Kupffer's cell hyperplasia, and hepatocellular dysplasia. Specific findings suggestive of GVHD, including bile duct damage (i.e., ductopenia and dystrophia), cholangitis, nuclear pleomorphism, and epithelial cell dropout were also recorded.

Biochemical, Virological, and Immunologic Assays.

Routine biochemical liver function tests were performed systematically throughout the clinical course of all patients. Investigations of hepatitis A and E virus (HAV and HEV) antibodies (Abs) (i.e., immunoglobulin IgM), hepatitis B surface antigen, and Abs to hepatitis B virus (HBV) surface and core antigens were carried out on serum samples. A diagnosis of hepatitis C was based on serum positivity for Abs to hepatitis C virus (HCV) and HCV RNA. Markers for other types of viral hepatitis, such as CMV, Epstein Barr virus (EBV), and herpes simplex virus HSV1-2, were also tested. Human herpes virus HHV6 was detected by polymerase chain reaction (PCR) in the plasma and liver.

The presence in sera of autoimmune liver Abs, such as antinuclear Abs (ANA), anti–smooth muscle antigen (SMA), anti–liver-kidney microsome type 1 (LKM-1), antiliver cytosol type 1 (LC1), and antimitochondrial Abs (AMA), was investigated using indirect immunofluorescence (IIF) on frozen tissue sections of rat stomach, liver, and kidney.

Serological Proteomic Analysis.

Immunoreactivity of sera from 3 patients (P1, P2, and P3) was determined by two-dimensional (2D) immunoblotting before, and at the onset of, liver dysfunction. The immunoreactive spots of interest were identified by mass spectrometry (MS).

Antigen preparation from liver homogenates

All chemical reagents used were obtained from Sigma-Aldrich (St-Quentin, France), unless otherwise stated. Rat livers from male Wistar rats (Charles River, Saint Germain sur l'Arbresle, France) were homogenized in 10 mM of Tris, 250 mM of sucrose, and 1 mM of 4-(2-aminoethyl) benzenesulfonyl fluoride (AEBSF) buffer using a Potter-Elvehjem apparatus. Liver homogenates were then fractionated by differential centrifugation (as described elsewhere) to obtain mitochondrial, microsomal, and cytosolic fractions.14 Nuclear fractions were obtained after sucrose gradient density ultracentrifugation.15 All subcellular fractions were stored as aliquots at −80°C until use.

2D electrophoresis and immunoblotting

Fraction aliquots were solubilized in a buffer (7 M of urea, 2 M of thiourea, and 4% CHAPS; w/v) in the presence of Orange G and 0.5% immobilized pH gradient (IPG) buffer at pH 3-10 (GE Healthcare, Saclay, France). 20 mM of dithiothreitol (DTT) and 20 mM of AEBSF were added extemporaneously. For each fraction, proteins were applied to Immobiline DryStrip (13 cm, pH 3-10; GE Healthcare) at rates of 250 µg for future immunoblotting and 1 mg for future Coomassie blue staining. Isoelectric focusing was performed with a voltage that was gradually increased to reach 23,000 Vh. For subsequent immunoblotting, proteins (after equilibration) were first resolved on 10% polyacrylamide separating gels,16 transferred to nitrocellulose membranes in accord with Towbin's protocol,17 and then probed with sera collected before and at the time of onset of hepatic dysfunction (dilution 1:2,000) and then incubated with (1:3,000) diluted horseradish-peroxidase–conjugated antihuman Ig (Bio-Rad, Hercules, CA). Proteins were detected by chemiluminescence according to the manufacturer's instructions (ECL Plus Western Blotting Detection kit; GE Healthcare). After transfer, the resulting gels were silver-stained. For future protein digestion, 1-mg protein-loaded gels were stained with Coomassie blue. For each patient and each cellular fraction, the silver-stained transferred gels and immunoblottings were scanned and then superimposed using Adobe Photoshop software to detect spots that were only revealed by sera collected at the time of hepatic dysfunction. Spots of interest were then localized on the corresponding scans of Coomassie blue-stained gels.

Procedures for protein and peptide preparation

Briefly, the selected proteins were excised from the Coomassie blue–stained gels, washed in a mixture of 25 mM of ammonium bicarbonate and acetonitrile (J.T. Baker Chemicals B.V., Deventer, The Netherlands), reduced in 10 mM of DTT, and alkylated in 55 mM of iodoacetamide (Sigma Aldrich). They were digested overnight in gel with trypsin (sequencing grade modified trypsin; Promega, Madison, WI).11,18 Previous washing and digestion procedures were automated using a ProGest workstation (Genomic Solutions, Ann Arbor, MI). Peptides were extracted using a mixture of 60 parts acetonitrile, 40 parts ultrapure water, and 1 part formic acid (VWR, Fontenay-sous-Bois, France). Peptide extracts were dried in a Speedvac concentrator, solubilized in a 2% formic acid solution, and then sonicated.

Protein identification

Protein identification was achieved using tandem matrix-assisted laser desorption-ionization (MALDI) time-of-flight (TOF) MS and was confirmed by nano high-performance liquid chromatography (HPLC) coupled with an LTQ Orbitrap.


A solution of α-cyano-4-hydroxycinnamic acid (CHCA; 4 mg/mL in water), trifluoroacetic acid (TFA; 0.1%), and acetonitrile (50/50), was mixed with the solubilized peptide mixture and applied twice to an appropriate plate. Peptides were analyzed by MS/MS using a 4800 MALDI TOF/TOF analyzer (AB SCIEX, Les Ulis, France) calibrated with a standard mix of calibrants. Data mining was performed in the UniProtKB databank, using ProteinPilot software (AB SCIEX, Les Ulis, France).

Nano HPLC Coupled With an LTQ Orbitrap.

Peptide extracts were analyzed by nano HPLC U3000 coupled with an LTQ Orbitrap (Thermo Instruments, Les Ulis, France). The six most intense peptides were fragmented, and the MS1 spectra were acquired at a resolution of 60,000. Data mining was performed against the rat UniProtKB data bank, using Proteome Discoverer 1.1 software (Thermo Instruments), with an accuracy of less than 5 ppm for parent ions and 0.8 Da for fragments.

All the proteins thus identified were analyzed using Pantherd software to determine their gene ontology parameters.


Clinical, Biological, and Pathological Features of non-GVHD Hepatitis After BMT.

Biological and histological features of the patients at the diagnosis of acute hepatitis are reported in Table 1. Mean values for total bilirubin, gamma-glutamyl transferase (GGT), and aminoaspartate transferase (AST) levels, as well as the prothrombin time, were, respectively, 121 µmol/L (range, 29-270), 933 IU/L (range, 455-1,968), 1,438 IU/L (range, 538–2,900), and 74% (range, 37-100). IgG levels were high in P1 (24.5 g/L) and P5 (24.4 g/L), but normal in the other patients.

Pathological examination revealed features of acute hepatitis with interface (n = 4) and lobular (n = 4) necroinflammatory activity. An abundant inflammatory infiltrate, including plasmocytes, was present in three patients (P1, P3 and P5) (Fig. 1). During the initial presentation, fibrosis was mild or absent in P1, P3, P4 and P5, and advanced in P2. There was no evidence of pathological features of GVHD or veno-occlusive disease. Moreover, at the onset of liver dysfunction, no extrahepatic symptoms suggestive of GVHD could be detected.

Figure 1.

Histological fractions during the initial presentation in P1. Interface necroinflammatory activity with plasmocytes.

In the control groups, the histological pattern of acetaminophen hepatitis differed markedly from the pattern described above (Supporting Fig. 1). Necrosis was the sole feature observed, without any lymphoplasmocytic infiltrate.

With respect to autoantibody detection, no patient was positive for anti-SMA, anti-LKM1, or anti-LC1 before and at the onset of hepatic dysfunction. ANA were negative in all patients before hepatic disease and remained negative in P1, P4, and P5, although becoming positive in P2 and P3 (1:80 and 1:640, respectively). All viral markers tested, namely HAV, HBV, HCV, HEV, CMV, EBV, HHV6, and HSV, were negative in patients P2, P3, P4, and P5 before BMT and remained so after the onset of hepatic dysfunction. In P1, although the HCV test was positive before BMT, no HCV RNA could be detected by PCR.

One-Dimensional Reactivity Patterns

Immunoblottings performed on cellular fractions displayed very few common stained bands between patients P1-P3 and the two control groups (Supporting Fig. 2).

Figure 2.

Representative patterns of immunoreactive proteins recognized by patient sera. (A) Nuclear fraction. (B) Mitochondrial fraction. (C) Microsomal fraction. (D) Cytosolic fraction. Spots were more numerous and more intensely stained by serum collected at the onset of the hepatic dysfunction than before.

2D Reactivity Patterns Before and at the Diagnosis of Non-GVHD Hepatitis.

A comparison of 2D immunoblotting patterns showed that immunoreactive spots were more numerous and more intensely stained by the three sera collected at the onset of the hepatic dysfunction than by those collected before, regardless of the type of liver subfraction used as the antigen (Fig. 2). Moreover, a marked patient-related heterogeneity of the patterns was noted (Fig. 3). A total of 259 spots only present at the time of onset of liver dysfunction were detected (Supporting Fig. 3).

Figure 3.

Immunoblotting patterns displayed by the three patient sera with the microsomal fraction used as antigen. A marked patient-related heterogeneity of the patterns was noted with sera collected at the onset of the hepatic dysfunction.

Identification of Proteins Present at the Onset of Non-GVHD Hepatitis.

Spots that were only stained by sera at the onset of hepatic failure were excised and subjected to in-gel trypsin digestion. We identified 240 spots with a good correspondence between observed and theoretical MM and pI values, a significant score, and a suggestive combination of the number of matching peptides and percentage coverage (Supporting Table 2). These 240 identifications corresponded to 103 proteins. The presence of multiple isoforms of the same protein explained the discrepancy between the number of identified proteins and that of the spots detected.

Table 2. Twelve Common Proteins* Detected With Patient Sera (P1-P3), Collected at the Time of Hepatic Dysfunction
  1. a

    Most of those common proteins display catalytic activity.

60S acidic ribosomal protein P0Translational repressor
Arginase 1Catalytic activity in urea cycle
ATP synthase subunit alphaATP production
Carboxylesterase 3Metabolism of xenobiotics, natural substrates
CATCell protection from H2O2; growth of cells
Dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complexCatalytic activity
Hydroxy methyl glutaryl-CoA synthaseCholesterol, lipid, steroid biosynthesis
Long-chain–specific acyl-CoA dehydrogenaseFatty acid and lipid metabolism
Medium-chain–specific acyl-CoA dehydrogenaseFatty acid and lipid metabolism
Transitional endoplasmic reticulum ATPaseTransport
Ubiquinol cytochrome C reductase complex core protein 1Mitochondrial respiratory chain/proteolysis
Very-long-chain–specific acylCoA dehydrogenaseFatty acid and lipid metabolism

Genes encoding these proteins were analyzed using the Gene Ontology database (version 7.0; available at The terms “molecular function” and “biological process” were studied. Proteins involved in catalytic activity as a molecular function and a metabolic process as a biological function were dominant (Fig. 4).

Figure 4.

Assignment of gene ontology terms to the proteins identified.

Only 12 of the proteins identified in any cellular fraction were detected by all three patient sera (Table 2), namely 60S acidic ribosomal protein P0, arginase 1, adenosine triphosphate (ATP) synthase subunit alpha, carboxylesterase 3, catalase (CAT), pyruvate dehydrogenase complex, hydroxyl methyl glutaryl-CoA (coenzyme A) synthase, long-chain–specific acyl-CoA dehydrogenase, medium-chain–specific acyl-CoA dehydrogenase, transitional endoplasmic reticulum ATPase, ubiquinol cytochrome C complex core protein 1, and very-long-chain–specific acylCoA dehydrogenase.

Course of Liver Disease After the Diagnosis of Non-GVHD Hepatitis.

In all 5 patients diagnosed with non-GVHD hepatitis, immunosuppressive therapy with corticosteroids (n = 5) and cyclosporine (n = 2) was resumed. Within a mean period of 20 weeks after this resumption, their liver function parameters had normalized. Although the biological parameters improved in P1, the patient presented with ascites and edema. A second liver biopsy performed 6 weeks after the first revealed a marked reduction in inflammatory markers and extensive fibrosis (Fig. 5). Ascites was controlled with diuretic therapy and the liver parameters were still within the healthy range 6 months later. In the case of P5, corticosteroids were withdrawn 1 year after the episode of acute hepatitis, and a further episode of acute hepatitis occurred 4 years later. A new liver biopsy revealed interface and centrolobular necroinflammatory hepatitis with plasmocytes. A new course of corticosteroid therapy was initiated, and a normalization of liver function parameters was achieved rapidly. In P1-P4, very slow tapering of the corticosteroid therapy was pursued from 10 mg/day, with a reduction of approximately 1 mg every month. No recurrence of liver disease was observed in any of these patients (Fig. 6).

Figure 5.

Second liver biopsy performed 6 weeks after the first in P1. Reduction of inflammatory markers, but occurrence of extensive fibrosis.

Figure 6.

Treatment and course of liver transaminase, bilirubin, Ig levels, prothrombine time, and autoantibodies by IIF over 22 months after BMT.


The results reported in this study shed new light on the characterization of potentially severe non-GVHD hepatitis resembling AIH that occurs after BMT. The clinical features of the five cases described here were similar to six other case reports, presenting no history of liver toxic drug absorption, patent viral infection, or histopathological findings consistent with GVHD, but with features suggestive of AIH.5-10 BMT was well accepted by all the patients, as shown by the course of microchimerism tests during the year that followed transplantation. Indeed, chimerism levels in blood or bone marrow reached 100% donor cells in 4 patients within 6 months of BMT (data not shown). All but 2 of these patients developed a comparable clinical sequence of events. As in previous case reports,8,10 GVHD occurred during the first weeks or months after BMT, involving skin or gut expression. The patients were treated with increased levels of immunosuppressive therapy. In the 2 patients who did not present with GVHD, we cannot exclude the possibility of a GVHD without any clinical expression because of the immunosuppressive therapy. Overall, all the patients experienced acute hepatitis at the end of, or after, a reduction of immunosuppressive therapy.

Despite the observation of histological features of AIH, two major criteria for this disease were often absent in the cases reported here: hypergammaglobulinemia and the presence of autoantibodies usually found by routine IIF.19 One-dimensional immunoblotting patterns showed only a few common bands between P1, P2 and P3, and the control groups of AIH and acetaminophen hepatitis sera. Furthermore, histological features differed markedly from those observed in acetaminophen hepatitis20 and were not typical of the liver manifestations of GVHD.21,22

This is the first report of a comparison of immunoblotting patterns using chemiluminescence, a highly sensitive detection tool, which revealed the emergence of numerous autoantigens recognized by three patient sera contemporaneous with this non-GVHD hepatitis. Identification of these immunoreactive spots using MS indicated that 103 proteins became antigenic targets, of which only 12 were recognized by all three sera. As proposed by Mori et al.,6 the heterogeneity of the autoimmune response could be explained by GVHD-induced tissue damage. Indeed, the first hypothesis advanced suggests that bacterial products or virus crossing the damaged gut epithelial barrier during GVHD might induce the activation of immunity by Toll-like receptors (TLRs). Autoreactive lymphocytes may be present in the liver without developing an immune response,23 but TLR3 stimulation induces the production of proinflammatory cytokines and the development of autoimmune phenomena.

On the other hand, in accord with Teshima et al.,24 we can speculate that as a result of skin or gut damage, the patients in our study released modified or cryptic antigens that were not recognized as self, and were able to produce autoreactive cells.

Finally, because the recognition as “non-self” by the donor's immunocompetent cells affects all the recipient's tissues, damage might not be restricted to the skin and gut. In particular, the thymus epithelium might be altered,25,26 showing a depletion of thymocytes, a destruction of dendritic cells, a destruction of thymic epithelial cells, and a disappearance of Hassall's bodies after BMT. These alterations present no clinical translation, but can lead to either the production of autoreactive T cells, which are not destroyed during the selection process, or a deficiency in regulatory T cells specific to a self-peptide.

In our study, the autoantigen spread revealed by MS was compatible with a random destruction of tissues, thus explaining the appearance of numerous autoantibodies and the interindividual variations in the patterns observed. By contrast, in AIH, the number of autoantibodies is limited and the patterns are similar between patients. A study using serological proteome analysis performed by Xia et al.27 detected 14 antigenic targets in AIH patients, among which only four were also found in our study: fumarate hydratase; gamma actin; protein disulfide isomerase precursor; and alpha enolase.

Nevertheless, we identified 12 immunoreactive proteins that were common to the 3 patients in the context of liver failure. Some of them have previously been described during autoimmune processes, including 60S ribosomal protein P0 as an autoantibody target in systemic lupus erythematosus, the pyruvate dehydrogenase complex and transitional endoplasmic reticulum ATPase in primary biliary cirrhosis, and arginase 1, CAT, and transitional endoplasmic reticulum ATPase in AIH.28-32

The other information supplied by identification of these 12 common antigens was that many of them had previously been detected during several studies of the cell-surface proteome, such as ubiquinol cytochrome C reductase, CAT, transitional endoplasmic reticulum ATPase, arginase 1, and aldhehyde dehydrogenase.33,34

Last, but not least, another lesson learnt from this MS identification was the presence among the immunoreactive spots determined at the onset of hepatic dysfunction of proteins with a potential plasma membrane location, previously reported to be antigenic targets in AIH and, namely, cytokeratin 8 and 18, heat shock proteins HSP60, HSP70, and HSP90, transitional endoplasmic reticulum ATPase, and liver arginase.13 This observation raises the question of the active participation of these antigens in hepatocyte destruction. Indeed, it has been described elsewhere that autoantibodies to liver arginase display Ab-dependent cell-mediated cytotoxicity as well as direct cytotoxicity.35

To our knowledge, this study constitutes the most important collection of data on non-GVHD hepatitis mimicking AIH occurring after BMT. Its clinical and biological findings were in accord with previous case reports. All these reports5-10 had highlighted the role of GVHD in the pathogenic process, causing the transformation of an alloimmune process into an autoimmune reaction. In particular, the role of putative plasma membrane autoantigens in liver destruction needs to be further investigated. The identification of antibody targets by MS also showed that this liver disorder differs from de novo AIH occurring after liver transplantation.36,37

In conclusion, we suggest that any reduction in IS should be performed with caution, and all liver function parameters should be monitored closely after the withdrawal of IS after BMT and GVHD.