Plasma cell hepatitis in liver allografts: Variant of rejection or autoimmune hepatitis?


  • Anthony J. Demetris,

    Corresponding author
    1. University of Pittsburgh Medical Center, Pittsburgh, PA
    2. AP-HP Paul Brousse Hospital, Inserm U785, Villejuif, France
    • University of Pittsburgh Medical Center Montefiore, Room E741, 3459 5th Avenue, Pittsburgh, PA 15213
    Search for more papers by this author
    • Telephone: 412-647-2067; FAX: 412-624-6666

  • Mylene Sebagh

    1. AP-HP Paul Brousse Hospital, Inserm U785, Villejuif, France
    Search for more papers by this author

  • See Article on Page 861

In this issue, Fiel et al.1 use the phrase “plasma cell hepatitis (de novo autoimmune hepatitis)” to describe what they eventually conclude to be a variant of acute rejection. This study was likely prompted by curiosity about the significance of plasma cell–rich infiltrates in liver allograft biopsies. Patients were first chosen for study with a free-text search of pathology reports for the key words “liver allograft,” “lymphoplasmacytic,” and “plasma cell(s).” These results were then cross-referenced with all patients undergoing liver transplantation for hepatitis C virus (HCV)–induced cirrhosis. “Hepatitis” was not included in the search.


AIH, autoimmune hepatitis; HCV, hepatitis C virus; PCH, plasma cell hepatitis.

Twenty-eight (42%) initially selected patients were excluded because they had histological and/or clinical evidence of concomitant acute cellular rejection, chronic rejection, mechanical bile duct obstruction, hepatic artery thrombosis/stenosis, cytomegalovirus or other nonhepatotropic viral infection, or medication-induced hepatotoxicity. Additional patients were excluded because their biopsies contained ≤30% plasma cells.

The outcome of these highly selected patients was not good. Twenty-three of 38 (60%) died (n = 10) or developed graft failure and portal hypertension. Neither detailed descriptions of the causes of death and allograft failures nor the treatment approaches before these endpoints were included. One might assume that portal hypertension developed because of plasma cell hepatitis (PCH)–induced classical cirrhosis, but silently developing veno-occlusive disease due to centrilobular-based rejection can lead to the same endpoint.2 Also, lowering immunosuppression after widespread perivenular inflammation is found on biopsy might lead to an adverse outcome.

The authors concluded that PCH represents a form of acute rejection because this cohort (1) frequently developed PCH in association with suboptimal immunosuppression, (2) had a high incidence of acute rejection before the development of PCH that signaled a propensity toward rejection, and (3) had a better outcome when they were treated with increased immunosuppression.

Fiel et al.1 address 3 distinct but related questions in this article: (1) how does one distinguish between rejection and recurrent HCV, (2) how does one distinguish between HCV and autoimmune hepatitis (AIH), and, most importantly, (3) how does one distinguish between rejection and AIH? The Mt. Sinai group is to be congratulated on tackling some of these very difficult issues.

The authors appear to have distinguished recurrent HCV from acute rejection by examining the centrilobular areas for necroinflammatory activity involving a majority of perivenular regions. They stated the following:

We did not examine the significance of portal inflammation, interface hepatitis or lobular activity on clinical outcome or histological resolution because we could not reliably differentiate PCH from recurrent HCV based solely on these features. However, 25 patients (76%) had some degree of plasma cell infiltration pre-PCH with over half (14/25) already manifesting focal centrilobular necrosis (severe plasma cell infiltrate was found in two other patients but they also had ACR).

Indeed, centrilobular necroinflammatory activity was present in the index biopsy, which established the diagnosis of PCH, in all selected patients (100%).

Centrilobular necroinflammatory activity was noted to be a feature of acute rejection first nearly 40 years ago3 and then again in the mid 1980s when liver transplantation became more widespread.4 The subject arose once more in the early 1990s during preclinical and early clinical development of tacrolimus because an adverse event was raised as a potential cause of centrilobular necrosis and inflammation. After extensive study, however, the conclusion was that centrilobular inflammation and necrosis represented a manifestation of rejection,5 as had been concluded before.3, 4, 6

In 1993, after introducing tacrolimus at their center, the group then at Mt. Sinai,7 which included a few of the same authors involved in the study by Fiel et al.,1 concluded the following:

It is suggested that the development of perivenular injury, which is seen frequently in allograft biopsy specimens from patients on FK506 obtained at various intervals after transplantation, may be related to drug toxicity rather than to the process of allograft rejection.

The photomicrograph illustrating centrilobular FK toxicity in 19937 is remarkably similar to the one from Fiel et al. illustrating PCH with centrilobular rejection in 2008 (Fig. 1).

Figure 1.

The left panel illustrates a centrilobular lesion attributed to tacrolimus toxicity in 1993 by Hytiroglou et al. (Reprinted with permission from Transplantation.7 Copyright 1993, Transplantation Society.) In the right panel, Fiel et al. attribute a nearly identical lesion to acute rejection. (Reprinted with permission from Liver Transplantation.1 Copyright 2008, American Association for the Study of Liver Diseases.)

This publication7 contributed significantly to the use of the phrase “FK lesion” to describe centrilobular necrosis and inflammation in liver allografts. Some patients were treated with decreased immunosuppression, even though it had been described as a histological manifestation of acute rejection.3, 5, 6 A controversy subsequently arose about the cause of centrilobular lesions in liver allografts that prompted a number of articles on the subject.2, 8–19 All agreed that rejection was the most likely cause,2, 8–19 including a study carried out prospectively in HCV+ recipients that used the following criteria for rejection in the context of recurrent HCV: “portal inflammation with inflammatory bile duct damage involving ≥ 50% of the bile ducts” or “mononuclear perivenular inflammation involving ≥ 50% of the terminal hepatic venules, associated with hepatocyte necrosis and/or dropout.”20

If some of the patients with centrilobular necrosis and inflammation in the study of Fiel et al.1 were not treated with increased immunosuppression because of assumed tacrolimus toxicity, the poor outcome might be at least partially attributable to a center-specific “nontreatment” effect. Although details are not given, presumably the distinction between PCH and conventional or typical acute rejection was based on the presence of ≥30% plasma cells in the infiltrates and the prevalence of bile duct damage. Regardless, the authors make a convincing story that the centrilobular changes are not related to HCV-induced hepatitis alone. Instead, the centrilobular changes more likely represent an immunosuppression-responsive immune reaction, which is consistent with the numerous previous studies already mentioned.

Prominent plasma cell infiltrates and interface activity are reasonably good markers of AIH in native livers,21, 22 and this led Fiel et al.1 to consider whether PCH might represent de novo AIH. Plasma cell–rich liver infiltrates are also seen in hepatitis A and cytomegalovirus hepatitis, which Fiel et al. excluded. They eventually reasoned, however, that PCH was not true AIH because (1) AIH rarely occurs before 2 years after transplantation; (2) rapid resolution of plasma cell infiltrates after increased immunosuppression is not typical of AIH, and (3) only 14 of 23 patients had any evidence of autoantibodies, and these were at low titers (generally ≤1:40). Some might question their reasoning, however, because liver allograft recipients already receive potent baseline immunosuppression. Also, autoantibody titers might be lower and liver infiltrates might be more steroid-responsive than in the non-transparent setting. In addition, other surrogate markers of autoimmunity, such as hypergammaglobulinemia, were not measured.

Khettry et al.23 published a study very similar to that of Fiel et al.1 They followed the course of HCV+ recipients in their program. Nine of 92 (10%) developed what they termed “AIH-like” HCV based on histologic hepatitis with moderate to severe portal, periportal, and lobular necroinflammation with prominent plasma cells. These patients also showed a significantly higher incidence of central perivenulitis, but in contrast to Fiel et al., precise criteria were not given for either “prominent plasma cells” or the prevalence of central perivenulitis (that is, the percentage of central veins involved). The studies, therefore, cannot be compared directly. Regardless, serologic evidence of autoimmunity, either autoantibodies and/or hypergammaglobulinemia, was present in 6 of 9 (66%) of the AIH-like HCV patients. This supports the notion that prominent plasma cells are a marker of autoimmunity in liver allograft biopsies with chronic hepatitis, as in other settings.21, 22

Similarly to the study of Fiel et al.1 the outcome of liver allograft recipients with plasma cell–rich AIH-like HCV was also not good in Khettry et al.'s study.23 However, treatment details about immunosuppression management before and after the diagnosis of AIH-like HCV and after transplantation were not included in any detail. One patient died, and 7 developed significant and progressive fibrosis, presumably the type associated with chronic hepatitis and not the type associated with central vein inflammation. Interestingly, these 9 patients were also more likely to show plasmacytic periseptitis in the native liver and central perivenulitis after transplantation than patients showing other patterns of recurrent HCV.23 On the basis of this experience, they concluded that “the AIH-like group in our study represents a subtype of recurrent HCV with possibly an altered immune reaction on the part of the host.”23 Presumably, they are referring to AIH-like HCV as seen in the general population.21, 22

Yet another study on the same topic by Berardi et al.24 reported a series of 9 of 44 HCV liver allograft recipients who developed what they termed “de novo autoimmune hepatitis” after being treated with pegylated interferon alpha-2b and ribavirin for at least 6 months for HCV recurrence. These patients developed significant graft dysfunction and hepatitis despite HCV-RNA clearance in all but 1 case. Laboratory, microbiological, imaging, and histological evaluations led them to a diagnosis of de novo AIH according to clinicopathologic criteria defined by the International Autoimmune Hepatitis scoring system.25 Three patients developed other definite autoimmune disorders (overlapping antimitochondrial antibody–positive cholangitis, autoimmune thyroiditis, and systemic lupus erythematosus, respectively). Withdrawal of antiviral treatment and treatment with prednisone resulted in different outcomes (5 remissions and 4 graft failures with 2 deaths). In this study, the contention that the liver allograft damage was related to autoimmunity was more convincing because HCV had presumably been cleared, but it is uncertain whether polymerase chain reaction testing was carried out on liver tissue to detect latent HCV infection.

All of the aforementioned studies1, 23, 24 show that liver allograft recipients with HCV can develop autoimmune phenomena similar to those observed in HCV+ patients from the general population.26 It is well known that chronic HCV infection alone can be associated with multiple immune-mediated extrahepatic manifestations such as mixed cryoglobulinemia, Sjögren-like syndrome, the presence of serum rheumatoid factor, the production of autoantibodies, B-cell non-Hodgkin lymphoma, and liver inflammation resembling AIH (reviewed in Czaja and Carpenter21 and Kessel and Toubi26). Also, antiviral therapy with interferon can further increase the risk of emergent autoimmune phenomena.24, 27 HCV lymphotropism and enhanced T cell apoptosis likely contribute to viral persistency and disease severity.26 Molecular mimicry, defects in dendritic cell maturation, enhanced regulatory cell number and function, and the resistance of the CD5+ B-cell subpopulation to apoptosis probably all contribute to the autoimmune phenomena.26

In the general population, recognition of AIH-like features in chronic HCV is reliably based on the presence of significant plasma cells in the biopsy and aggressive interface activity.21, 22 However, the correlation is not perfect, and reproducible guidelines on what constitutes a plasma cell–rich infiltrate are not available. Regardless, the AIH-like pattern of HCV-induced liver injury in the general population has been associated with higher serum levels of gamma-globulin and immunoglobulin G, a greater frequency of cirrhosis, a higher mean Knodell score, a higher frequency of human leukocyte antigen DR3, and a greater occurrence of high-titer smooth muscle antibodies.21 This suggests that in some patients, HCV infection triggers a genetic susceptibility to autoimmune phenomena that also contributes to the tissue damage. Alternatively, an initially antiviral immune response might eventually spread to include autoepitopes.28, 29

Immunosuppressive therapy can lessen the severity of liver damage in patients with HCV showing AIH-like features in the general population.30 This occurs, however, at the expense of enhancing HCV replication and impeding eventual HCV clearance.30 It has been suggested, therefore, that the treatment of patients with HCV and evidence of immunological disorders should be based on the predominant manifestations.31 If autoimmune features predominate (for example, antinuclear or smooth muscle antibodies, Liver-kidney microsomal antibodies-1 > 1:320, active concurrent autoimmune disease, and clinical, laboratory, and histologic features of AIH) or autoimmune-like tissue damage is severe, then corticosteroids should be considered.30, 31 In contrast, interferon should be given when HCV or foreign antigen-driven disease predominates [antinuclear or smooth muscle antibodies < 1:320, anti-Liver kidney microsomal-1 positivity, ancient or inactive concurrent immune disease, very few or no criteria for AIH, cryoglobulinemia (symptomatic), glomerulonephritis, cutaneous vasculitis, and systemic vasculitis].31

Therefore, HCV+ liver allograft recipients are quite similar to HCV+ patients in the general population, except that liver allograft recipients can also develop rejection. This further complicates the situation. Accordingly, a subpopulation of HCV+ liver allograft recipients will develop AIH-like features, as shown by Khettry et al.23 Also, these patients will likely experience more aggressive disease similar to that of patients with AIH-like HCV in the general population.21 Another subpopulation of HCV+ liver allograft recipients will develop significant acute and even chronic rejection, as shown by Fiel et al.1

It is still uncertain whether plasma cell–rich centrilobular lesions in HCV+ liver allograft recipients represent rejection, as proposed by Fiel et al.,1 or an altered immune variant (autoimmune?) of HCV, as suggested by Khettry et al.23 The uncertainty arises from the fact that we do not know how to distinguish alloreactivity from autoreactivity. Incomplete data from both studies and the need for more study make it difficult to endorse either viewpoint. In Fiel et al.'s study, strict criteria for AIH might not be fulfilled completely because liver allograft recipients were already maintained on heavy immunosuppression, gamma globulin levels were not reported, and markers of autoimmunity were not routinely measured. Conversely, in Khettry et al.'s study, 1) not all cases of AIH-like HCV showed central perivenulitis; 2) the prevalence, or percentage of central veins involved with central perivenulitis in each biopsy was not mentioned; 3) no precise definition of plasma cell prominence was given; and 4) autoantibodies, which were not measured routinely in all groups, are frequently detected in otherwise typical acute and chronic rejection.32, 33 It would be difficult, therefore, to exactly repeat this study in another center.

Regardless, perivenular inflammation and necrosis involving a majority of central veins, plasma cell–rich or not, are features of both acute rejection and AIH.34 Also, these changes are responsive to increased immunosuppression, regardless of whether the patient is HCV+ or not and regardless of whether the patient is an allograft recipient or not.1, 20, 23, 24, 30 Other causes of perivenular necrosis and inflammation should also be considered, but they are much less likely than rejection and AIH.35 In contrast, perivenular inflammation and necrosis involving a majority of central veins are not common features of the majority of recurrent HCV cases in allografts or of HCV cases in native livers, unless they are accompanied by a significant autoimmune/rejection reaction. Also, whether one decides to treat rejection or autoimmune phenomena in HCV+ patients depends on the severity of the perivenular damage and whether the short-term gain of less liver damage offsets a diminished ability to eventually clear the HCV infection.

In our experience, left untreated, significant (moderate or severe) perivenular damage,34 whether plasma cell–rich or not, hastens the development of portal-to-central and central-to-central bridging fibrosis, as reported in all of the aforementioned studies.1, 23, 24 Also, significant rejection and autoimmune phenomena often occur coincident with a significant anti-HCV immune response and relatively low or absent circulating HCV-RNA levels;20 this is similar to the findings of Berardi et al.24

The final and most difficult question to answer is then this: how do we distinguish AIH from rejection? The proportion of plasma cells in the infiltrates is currently used as an important, albeit imprecise, marker. Also included are the prevalence of bile duct damage and the severity of the interface activity. Rejection-related infiltrates usually contain less than 30% plasma cells; lymphocytic damage involves a majority of bile ducts, and interface activity, if present, is usually mild.34 In contrast, AIH usually shows more severe interface activity and less bile duct damage, and plasma cells compose a greater percentage of the infiltrate than in rejection.34 However, no precise cutoff for the percentage of plasma cells has been established, and none of these criteria have been tested prospectively for reproducibility, predictive value, or an association with serologic evidence of autoimmunity.

Separating rejection from autoimmunity long after transplantation might be an exercise in semantics because epitope spreading occurs with all immune responses.28, 36–39 This refers to the process by which initially there is a narrow range of antigens recognized early in an immune response. Later, however, tissue damage uncovers additional epitopes (autoantigens or other foreign non–major histocompatibility complex antigens) that continue to drive the response,28, 36–39 and chronic inflammatory disorders frequently contain plasma cells.40–43

Acute rejection should be diagnosed when the immune response is directed primarily at antigens unique to the allograft liver,39, 44 such as mismatched major histocompatibility antigens or polymorphic drug-metabolizing isoenzymes.45, 46 Conversely, AIH should be diagnosed when the target antigen is shared exactly by the donor and recipient and tissue damage is mediated by memory cells in the recipient.44

Because we cannot yet determine with certainty the allospecificity or autospecificity of a liver-based immune response, we rely on surrogate markers, such as biopsy evidence of significant necroinflammatory interface activity, plasma cell prominence, hypergammaglobulinemia, autoantibodies, and steroid dependence.47 In addition, we do not know if rejection targeting non–major histocompatibility complex antigens appears histopathologically similar to rejection targeting other mismatched polymorphic proteins unique to the allograft.45 Some evidence suggests that it might differ.

Therefore, because criteria are imprecise, we do not know how to categorize some patients with intrahepatic, nonviral immune responses and whether any of the parameters measured affect prognosis or responsiveness to therapy. Berardi et al.24 used the International Autoimmune Hepatitis Group system, which is a reasonable approach. However, several criteria in that system are probably not relevant in the liver allograft setting.44, 48 Perhaps transplant-specific revisions to the International Autoimmune Hepatitis Group system are needed.

Most programs, however, do not routinely screen liver allograft recipients for autoantibody production, gamma globulin levels, or cryoglobulins. Also, some programs do not routinely conduct human leukocyte antigen typing, cross-matching, or serological screening for anti-donor antibodies. Without at least some of these data, liver allograft recipients will be difficult to classify further. To understand the significance of plasma cell–rich infiltrates in liver allograft biopsies, it is important to understand plasma cell biology in general.

Plasma cells mature from resting B cells after antigen stimulation, usually in the lymph nodes.49 Most plasma cells rapidly die, however, after maturation. This occurs as a result of the stress on the endoplasmic reticulum caused by massive production of antibodies, unless an unfolded-protein response, which is protective, is induced by external survival signals.42, 43 These survival signals, or plasma cell niches, can be located in secondary lymphoid organs, inflamed tissue, or, most commonly, bone marrow.42, 43 Arrival at these niches can lead to long-lived plasma cells,42, 43 which produce up to 40% of autoantibodies in experimental animals.41 Whether some of the plasma cell–rich infiltrates in autoimmune liver disease are long-lived is uncertain.

Plasma cell precursors, known as plasmablasts, migrate to these niches. Interferon-γ, a hallmark of inflamed tissue, increases both the expression of the chemokine receptor CXCR3 by B cells differentiating into plasmablasts and the expression of CXCR3 ligands by cells present in inflamed tissue.42, 43 To survive, plasma cells require both specific external signals and a molecular competence to respond to these signals. This depends on the expression of the transcription factor B-lymphocyte–induced maturation protein 1,42, 43 which protects the plasma cell from death.42, 43 Long-lived plasma cell survival also depends on the nuclear factor Aiolos because they cannot be derived from Aiolos-deficient B cells.42, 43 Bone marrow–derived plasma cell survival factors in vitro include CXCL12 (the ligand for CXCR4), tumor necrosis factor, interleukin-5, interleukin-6, and ligands for CD44 (such as hyaluronic acid).

Finally, comparing the results of these liver studies1, 23, 24 to studies of plasma cell–rich infiltrates in other solid organ allografts provides a broader perspective. In kidney allografts, plasma cell–rich rejection usually occurs late after transplantation and is associated with the production of anti-donor antibodies and poor prognosis.50–53 Plasma cell–rich infiltrates can also be seen with polyoma viral nephritis54, 55 and during rejection in patients infected with other viruses.56 Even though autoantibody production can be detected in graft-infiltrating lymphocytes,51, 57 there is little or no discussion of “autoimmune nephritis” in the kidney allograft literature. Instead, it seems to be accepted that epitope spreading results in a slightly different histopathologic appearance between early and late acute rejection and that plasma cells are a common component of chronic inflammation and are usually associated with a degraded prognosis, as in all 3 liver studies discussed previously.1, 23, 24