“Plasma Cell Hepatitis” in Liver Allografts: Identification and Characterization of an IgG4-Rich Cohort

Authors

  • M. Castillo-Rama,

    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Department of Pathology, Division of Liver and Transplantation Pathology, Thomas E Starzl Transplantation, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • M. Sebagh,

    1. Laboratoire d'Anatomie Pathologique, AP-HP Hôpital Paul Brousse, Villejuif, France
    Search for more papers by this author
  • E. Sasatomi,

    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Department of Pathology, Division of Liver and Transplantation Pathology, Thomas E Starzl Transplantation, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • P. Randhawa,

    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • K. Isse,

    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Department of Pathology, Division of Liver and Transplantation Pathology, Thomas E Starzl Transplantation, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • A. D. Salgarkar,

    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • K. Ruppert,

    1. Graduate School of Public Health, Department of Epidemiology, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • A. Humar,

    1. Department of Surgery, University of Pittsburgh Medical Center, Pittsburgh, PA
    Search for more papers by this author
  • A. J. Demetris

    Corresponding author
    1. Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA
    2. Department of Pathology, Division of Liver and Transplantation Pathology, Thomas E Starzl Transplantation, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author

Abstract

Plasma cell hepatitis (PCH), also known as “de novo autoimmune” hepatitis, is an increasingly recognized, but suboptimally named and poorly understood, category of late allograft dysfunction strongly resembling autoimmune hepatitis (AIH): They share plasma-cell-rich necro-inflammatory activity on biopsy, autoantibodies and steroid responsiveness, but overlap with rejection is problematic. A retrospective study of clinical, serological, histopathological and IgG4 immunohistological features of PCH (n = 20) in liver allograft recipients, native liver AIH (n = 19) and plasma-cell-rich renal allograft rejection (n = 20) showed: (1) high frequency (44%) of HLA-DR15; (2) less female predominance (p = 0.03) and (3) n = 9/20 PCH recipients showed >25 IgG4+ plasma cells/high-power field (IgG4+ PCH) versus AIH (n = 1/19, p = 0.008) or plasma-cell-rich kidney rejection (n = 2/20, p = 0.03). The IgG4+ PCH (n = 9) subgroup showed lower alanine transaminase (ALT) (p < 0.01) and aspartate transaminase (AST) (p < 0.05) at index biopsy but (a) higher plasma cell number/percentage, (b) more aggressive-appearing portal/periportal and perivenular necro-inflammatory activity and (c) more severe portal/periportal fibrosis than IgG4− PCH (n = 11). Significant demographic, histopathologic and plasma cell phenotype differences between PCH and AIH suggest distinct pathogenic mechanisms for at least the IgG4+ PCH subgroup likely representing an overlap between allo- and auto-immunity. IgG4+ PCH was associated with fibrosis, but also highly responsive to increased immunosuppression.

Abbreviations
AIH

autoimmune hepatitis

GITR

glucocorticoid-induced tumor-necrosis-factor-receptor-related protein

HAI

hepatitis activity index

HPF

high-power field

IgG4

immunoglobulin G4

PCH

plasma cell hepatitis

Qdot

quantum dot

RAI

rejection activity index

Introduction

Plasma cell infiltrates in kidney allografts are usually associated with viral infections, acute rejection with a poor prognosis and chronic rejection [1, 2]. In liver allografts, plasma-cell-rich infiltrates associated with a “hepatitic” pattern of injury have been referred to as de novo “autoimmune” hepatitis or plasma cell hepatitis (PCH) [1]. This increasingly recognized, but suboptimally named and poorly understood, category of late allograft dysfunction affects patients transplanted for nonimmune disorders. Clinically, histopathologically and serologically, however, PCH does indeed resemble autoimmune hepatitis (AIH) sharing plasma-cell-rich necro-inflammatory activity on biopsy, autoantibody production and steroid-responsiveness [3-5], but overlap with rejection is problematic [1, 6].

The concept of allograft “autoimmunity” is confusing and diagnosing PCH can be challenging because of clinical, serological and histological features that overlap with rejection [7-9]. It is also difficult to predict in whom it will occur and who will respond to increased immunosuppression or corticosteroid treatment (e.g. co-existing hepatitis C virus [HCV] infection). More broadly, the pathophysiologic significance of plasma-cell-rich infiltrates in organ allografts is poorly understood despite the negative prognostic implications of their presence.

Immunoglobulin G4 (IgG4), the least plentiful IgG subtype, which can be produced by B regulatory cells [10], has traditionally been considered an anti-inflammatory immunoglobulin, capable of modulating the immune system [11, 12]. IgG4 antibodies, however, have also been implicated in the pathogenesis of autoimmune diseases. IgG4 production has been linked to high antigen concentration and chronic or prolonged antigen exposure and seems to play a role in inducing tolerance in allergen-specific immunotherapy [13]. The IgG4 constant domain has been used in the development of monoclonal antibodies (e.g. Natalizumab, Eculizumab) where effector function is undesirable [14].

Possible pathogenic mechanisms of PCH after liver transplantation were investigated by conducting a retrospective study of affected patients to determine the number/percentage of IgG4+ plasma cells within the plasma cell infiltrates, as well as their possible association with FOXP3+ and CD4+/FOXP3+ Treg cells, which might be involved in pathogenesis [15]. A small previous Japanese study failed to show any increased or preferential IgG4 usage in PCH after living donor liver transplantation [16]. Our results were compared with native liver AIH and kidney allograft biopsies with plasma-cell-rich rejection. A substantial subpopulation of liver allograft recipients with otherwise typical PCH by routine light microscopy show IgG4+-rich plasma cell infiltrates that were not detected at the same frequency in native livers with AIH or in kidney allograft recipients with plasma-cell-rich rejection. The clinical, biochemical, routine histological, IgG4+ and CD4+/FOXP3+ immunostaining features of these patients are reported, in detail, and discussed.

Materials and Methods

Patients

Liver allograft recipients at the University of Pittsburgh Medical Center (1995–2008; IRB# PRO10110393) were extracted from the EDIT database based on a histopathological-coded diagnosis of “plasma-cell-rich hepatitis (PCH)” or “autoimmune-like hepatitis,” which in turn, was based primarily on the histopathologic findings, as follows. PCH cases showed a “hepatitic” pattern of injury, characterized by mononuclear portal inflammation with necro-inflammatory-type interface activity where plasma cells comprised an estimated ≥30% of the total infiltrate, comparable to native livers with AIH [6]. Patients who underwent transplantation for AIH, as well as those with posttransplant biliary strictures or other technical complications that might contribute to an unexplained chronic hepatitis in the histopathology, were excluded. All patients included in this study were retrospectively identified based on the clinicopathological and serological profile, including a response to therapy. These patients were also included in a study of the routine histopathologic features of PCH [17].

Two “control” groups were included: (1) patients evaluated for new-onset AIH (native liver AIH) were randomly selected based on having noticeable plasma cells within the infiltrate prompting an IgG4 immunohistochemical stain as part of their diagnostic workup (n = 20); and (2) a group of renal transplant recipients with plasma-cell-rich rejection (n = 20) defined as comprising >30% of the infiltrates generally comparable to PCH. Three patients in the native liver AIH cohort had clinicopathological features of AIH/PBC overlap. One patient with clinicopathological evidence of AIH/PSC overlap syndrome was excluded from the final analysis. The purpose of the control groups was to determine whether IgG4+ plasma cells in the PCH group were (a) a consequence of alloimmunity or immunosuppression or (b) a liver context-specific phenomenon.

Immunohistochemistry

Deparaffinized and rehydrated liver biopsy sections were blocked for 6 min in a 3% hydrogen peroxide solution, and then steamed in target retrieval EDTA (pH 8; Life Technologies, Grand Island, NY) for 30 min after which they were washed in PBS. Protein block (Dako, Carpinteria, CA) was applied for 10 min, and slides were incubated at room temperature for 1 h with an IgG4 antibody (mouse monoclonal antibody to human IgG4; 1:1000; clone HP6205, Zymed, Camarillo, CA). They were then washed in PBS and incubated for 30 min with a polymer detection reagent (ImmPRESS Universal Reagent, Vector Laboratories, Burlingame, CA) followed by washes with PBS. The slides were developed by incubating for 10 min with a DAB substrate/chromogen solution (Vector Laboratories, Burlingame, CA), and then counterstained with Shandon hematoxylin (ThermoScientific, Kalamazoo, MI).

Multiplex quantum dot immunohistochemistry

All samples were sequentially stained with an unconjugated primary antibody for CD4 (Dako, clone 4B12, 1:40, Carpinteria, CA) followed by FOXP3 (Abcam, clone 236A/E7, 1:70, Cambridge, MA) and optimal biotinylated secondary antibodies followed by streptavidin-conjugated quantum dot (Qdot) nanoparticles: Qdot705 and Qdot605 (both at 1:50 dilution, Invitrogen, Eugene, OR), as described [18].

High-resolution whole-slide imaging

Slides were scanned on a Zeiss MIRAX MIDI whole-slide image (WSI) scanner (Carl Zeiss, Jena, Germany) and quantified using in-house developed IAE-NearCYTE WSI Analysis software [19]. Images correspond to WSI scans generated using a 40× objective (0.95 NA) with resultant image resolution of 0.116 microns/pixel2.

Histopathological analysis

A senior pathologist (A.J.D.), unaware of any clinical, serological or IgG4 staining results, performed a complete retrospective re-review of native liver AIH, PCH and kidney allograft cases for the purpose of this study and scored the histopathological parameters depicted in Table S1 based on current guidelines [6, 20-22]. The percentage of plasma cells, as percentage of the total infiltrate, in the centrilobular and portal/periportal infiltrates was estimated, separately, to the nearest 10%.

The absolute number of IgG4+ plasma cells was counted in one representative high-power (40× objective) field focused primarily on a portal tract, but occasionally captured some of the interface zone. This process was repeated for five separate portal tracts. The results were expressed as the mean of five portal tracts. The same process was carried out to determine the estimated ratio of IgG4+ cells/total plasma cells and, again, expressed as the mean of five portal tracts. Plasma cells were recognized by their distinctive nuclear morphology and peri-nuclear hof.

“Central perivenulitis” was graded according to Banff criteria for this entity [6]. The severity of fibrosis followed similar guidelines, previously outlined in the Banff consensus document on chronic rejection [21]. Specifically, “mild”: noticeable fibrous expansion of a majority of central veins; “moderate”: more striking expansion of a majority of central veins, as compared to mild, but without central-to-central, or occasionally, central-to-portal bridging; “severe”: as above for moderate, but also with evidence of bridging fibrosis.

All kidney allograft biopsies were also re-reviewed to ensure the presence of a dense interstitial infiltrate contained >30% plasma cells; the majority contained a higher percentage. The absolute number and ratio of IgG4+ plasma cell/total plasma cells were determined identically to the liver analysis, except that the high-power fields were selected from the renal cortical interstitium.

For the CD4+/FOXP3+ Qdot analysis, CD4+/FOXP3+ cells and CD4+FOXP3− cells were counted within portal tracts as previously described, and quantified manually using IAE-NearCYTE software tools. Analysis for the FOXP3+ single immunohistochemistry was performed in the same fashion.

The biopsy samples were also assessed for key morphological features associated with IgG4-related disease in the general population including: (1) dense lymphoplasmacytic infiltrates combined with a storiform pattern of fibrosis, (2) increased number of IgG4+ plasma cells, (3) obliterative phlebitis and (4) mild-to-moderate eosinophil infiltrates [23].

Statistical analysis

Data were shown as mean ± SD or median. Groups were compared using the Fisher test, and Wilcoxon rank-sum (Mann–Whitney) as appropriate. Exact Chi-squared analysis was performed to compare the proportion of patients, within each cohort, with IgG4 >25 cells/high-power field (HPF) and IgG4+ cells/plasma cells >25% (based on the sensitivity/specificity) to similar patients in the other cohorts. Analyses were performed using the SAS 9.2 software. Differences between groups were considered significant at p < 0.05.

Results

Clinical and IgG4 staining profile

The clinical and serological profiles of the 20 PCH patients showed a male:female ratio of 9:11 with a mean age of 57 ± 12 years (range: 23–80 years) whereas the ratio in the native liver AIH group was 1:18 (p = 0.03) with a mean age of 55 ± 12.9 (range: 19–71 years; p = NS vs. PCH age) (Table 1). The male:female ratio in patients with plasma-cell-rich kidney rejection was 12:8 with a mean age 49 ± 18 range: 18–78 years (p = 0.18). Ninety-five percent (n = 19/20) of PCH patients were on tacrolimus (TAC)-based immunosuppression either in combination with mycophenolate mofetil (MMF) (15%) or alone (80%); the remaining patient was on no immunosuppression. None had an HLA-DR3 allele, and only three had an HLA-DR4 allele, but 44% of patients (7/16 typed for HLA) had an HLA-DR15 allele, which is about three times the frequency in North American Caucasians [24]. HLA-DR data were only available in one native AIH patient (HLA DR7, 11). Autoantibody evidence of autoimmunity and/or hypergammaglobulinemia was available for 10/20 (50%) of the PCH and 16/19 (84%) of native AIH patients. No posttransplant allo-antibody information was available, but 17/20 (85%) had pretransplant crossmatch data.

Table 1. General demographic, IgG4, baseline immunosuppression and treatment of PCH, LFTs at index biopsy, histological follow-up and outcome of patients diagnosed with PCH
PatientAge (recipient/donor)Sex (recipient/donor)DR3 (recipient/donor)DR4 (recipient/donor)Original diseaseIgG4 (average 5 portal tracts ± SD)%IgG4/plasma cells (average 5 portal tracts)IS regimenTreatmentALT/AST at index BxLFTs improve/normalHistopathology (follow-up biopsy)Outcome
  1. TAC, tacrolimus; CS, corticosteroids; MMF, mycophenolate mofetil; RAPA, rapamycin; incr, increase; IS, immunosuppression.
  2. Mean postdiagnosis follow-up time of entire cohort 2.6 years (range 1 week to 10 years).
155/45M/FNo/noNo/yesHBV + HCC0 (±0)0TACCS, TAC incr211/82YesMild ductular proliferation, mild portal lymphocytic inflammation and portal fibrosisAlive
264/NAF/MNA/noNA/noPBC0 (±0)0TACCS, TAC incr440/488YesAlive
360/47M/FNo/noNo/noETOH + HCC0 (±0)0TACCS, Campath, RAPA491/587YesBridging fibrosis with persistent plasma cells (>20%) and hemosiderosisDeceased (2 years)—bowel infarction
423/24F/MNA/noNA/noBA0 (±0)0TACCS1930/2207YesDeveloped ductopenia at 5 yearsDeceased (5 years)—graft failure
557/44F/FNo/noYes/yesNASH0 (±0)0TACCS, MMF, TAC incr435/325YesAlive
662/49F/FNo/noNo/noNASH65 (±37.52)32TACCS, MMF, TAC incr402/241YesDeceased (1 year)—hyperkalemia
756/35F/MNo/noNo/noHCV20 (±14.40)32TAC + MMFMMF, TAC incr235/169YesNo significant plasma cells, probably ACR in context of recurrent HCVDeceased (3 years)—acute or chronic renal failure
854/34F/FNo/noNo/noPBC0 (±0)0TACCS, TAC incr54/55YesPersistent inflammation hepatitis/recurrent PBC/rejectionAlive
942/52M/FNo/noNo/noA1ATD47 (±17.56)28TAC + MMFCS, TAC incr112/63YesGraft failed because of cirrhosis with persistent plasma cells >30% (3 years)Deceased (3 years)—cirrhosis
1069/73F/MNo/noNo/noHCV0 (±0)0TACTAC incr613/624YesInitial response, but then developed bridging fibrosis when TAC loweredDeceased (1 year)—recurrent cirrhosis
1163/39F/FNo/noYes/noHCV41.4 (±14.47)32TACCS, TAC incr185/270YesRespondedDeceased (1 year)—multiple organ failure
1243/54F/MNo/noNo/noHCV + HCC0.2 (±0.45)2NATAC incr278/397YesRespondedDeceased (1 year)—recurrent HCC
1357/37M/FNo/noNo/noNAFLD19.2 (±5.93)42TACCS, MMF378/327YesPatient died with functioning graftDeceased (1 year)—motor vehicle accident
1450/29F/MNo/yesNo/noPBC5.2 (±3.83)16TACCS, TAC incr696/688YesDeceased (5 years)—graft failure
1558/72F/FNo/noNo/noNASH180 (±67.08)74TACCS, MMF151/299YesAlive
1680/37M/FNo/noYes/yesA1ATD132 (±73.94)84TACCS62/106YesAlive
1763/39M/FNo/noNo/yesHCV36 (±9.91)36TAC + MMFTAC incr52/55YesAlive
1863/NAM/NANA/NANA/NAETOH58 (±22.53)64TACCS99/209YesAlive
1958/26M/MNo/noNo/noHCV29 (±6.7)18TACCS, TAC incr163/164YesAlive
2067/NAM/NANA/NANA/NAHCV109 (±27.47)81TACCS159/278YesAlive

Immunohistochemical evaluation showed that more patients/biopsies from PCH patients contained >25 IgG4+ plasma cells/per portal tract HPF (n = 9/20; M = 6, F = 3) than patients/biopsies from native liver AIH (n = 1/19, p = 0.008) or plasma-cell-rich kidney rejection cases (n = 2/20, p = 0.03). This was also observed when biopsies comprising >25% IgG4+ cells/plasma cells per HPF were compared between the PCH group (n = 10/20) and the native liver AIH cases (n = 1/19, p = 0.003) or the kidney rejection cases (n = 1/20, p = 0.003) (Figure 1). The results for AIH patients in this study were similar to a previous study [25]. Hereafter, the PCH liver allograft recipients will be broken down into two subgroups: (1) IgG4+ PCH (>25 IgG4+ plasma cells/per portal tract HPF; n = 9; M = 6; F = 3) and (2) IgG4− PCH (<25 IgG4+ plasma cells/per portal tract HPF; n = 11; M = 3; F = 8). IgG4+ PCH status showed a nonsignificant trend toward male sex (6/9; 67%; p = 0.19) and increased age (mean age 62 ± 10, range: 42–80 years; p = 0.12), but there was no clear association with the underlying original disease or time elapsed since transplantation (Table 1). Seven patients had an HLA DR-15 allele, five of seven (5/7; 71.4%) showed IgG4+ PCH whereas the remaining two of seven (2/7; 28.6%) showed IgG4− PCH.

Figure 1.

Absolute number of IgG4+ plasma cells per high-power field (filled squares; left Y-axis) and estimated percentage of IgG4+ plasma cells/total plasma cells (open circles; right Y-axis) comparing plasma-cell-rich kidney allograft rejection (left grouping), native liver autoimmune hepatitis (middle grouping) and plasma cell hepatitis in liver allografts (right grouping). Despite an overall similarity in the severity of portal inflammation, as scored by the mHAI portal inflammation component score (data not shown), about half of the PCH patients (n = 9) showed >25 IgG4+ plasma cells per portal tract high-power field or IgG4+ plasma cells comprised >25% of the total plasma cells present (n = 10). Other histopathological differences, based on a “blinded” review of the slides, are shown in Table 3.

Four of the six (4/6; 67%) crossmatch positive recipients showed IgG4+ PCH, whereas only 3 of 11 (3/11; 27%) crossmatch negative recipients showed IgG4+ PCH (p = 0.11). The two patients (Patients 15 and 16) with the greatest number of IgG4 cells (132 and 180) showed serologic evidence of both autoimmunity and alloimmunity (Table 2). None of the PCH recipients (IgG4+ PCH or IgG4− PCH), however, had any record of the systemic manifestations associated with IgG4-related sclerosing disease [23].

Table 2. Serological studies in PCH patients and in those with native liver AIH
PatientAutoimmunityAlloimmunity
IgG (751–1560 mg/dL)ANA (<80)SMUSC (<20 EU)LKMA (0–20)MITOCH (0–20 EU)Crossmatch (CXM)CXM cell typeIgG4 cells/5 portal tracts
PCH patients
2A405050200PosB cell0
3A40505050NegT cell0
4A23500
11A401450NegT cell41.4
12A40NegT cell0.2
13A805050NegT cell19.2
14A40100NegB cell5.2
15A41501160PosT cell180
16A27008016401.8PosB cell132
18A8040.762.958
PatientAutoimmunityAlloimmunity
IgG (751–1560 mg/dL)ANA (<80)SMUSC (<20 EU)LKMA (0–20)MITOCH M2 (0–20 EU)Crossmatch (CXM)CXM cell typeIgG4 cells/5 portal tracts
  • IgG, immunoglobulin G; ANA, anti-nuclear antibody; SMUC, anti-smooth muscle antibody; LKMA, anti-liver/kidney microsomal antibody; MITOCH, anti-mitochondrial antibodies (M2).
  • 1Patient had clinicopathological evidence of AIH/PBC overlap.
  • 2Patient had a biopsy suggestive of AIH/PBC overlap, but this did not correlate with the clinical diagnosis.
Native AIH
1B2980<80108406
3B2358<80<20200.6
4B12578<808620500.2
5B2750<80100402.60.2
6B4930<801186.97.4
8B1000<403720200.2
9B21000<4037<20<2019.4
10B4000<80112<404.90.6
11B1128024143.93.4
12B918<807<403.90.8
14B997160511.8
16B1<40284037.10
17B160284011.90
18B20401280<20<20<200.2
19B1360<8076401.15.6
20B10106406401.70.8

Follow-up after a diagnosis of PCH showed that 10/20 (50%) patients died after a mean of 2.3 years (range 1–5 years) (Table 1). An extra-hepatic cause of death was suspected in six: bowel infarction (n = 1), hyperkalemia (n = 1), acute or chronic (Stage IV) renal failure (n = 1), multiple organ failure (n = 1), recurrent hepatocellular carcinoma (n = 1) and motor vehicle accident (n = 1), but three of these had either bridging fibrosis or early cirrhosis on follow-up biopsy. In the remaining four cases, extensive bridging fibrosis or established cirrhosis due to a combination of PCH and rejection (n = 3) or recurrent HCV (n = 1) was thought to have contributed to patient death. Three of nine (33%) IgG4+ PCH recipients versus 7/11 (64%) IgG4− PCH died (p = 0.37 Fisher's exact test). In the IgG4+ PCH group 67% (2/3 patients) died of an extra-hepatic cause compared to 57% (4/7 patients) in the IgG4− PCH group. The remaining IgG4+ PCH patient (1/3; 33%) and 3/7 IgG4− PCH patients (43%) died with bridging fibrosis or cirrhosis as described in Table 1.

Comparison of histological features in biopsies from native liver AIH and PCH recipients

Statistically significant differences between: (1) native AIH and PCH and (2) the IgG4− PCH and IgG4+ PCH cohorts are shown in Table 3. Significant differences in the first comparison included the Banff severity of perivenular inflammation or central perivenulitis [6] (p < 0.002), severity of perivenular fibrosis (p < 0.001), presence of centrilobular necrosis (p < 0.02), higher RAI component and total scores (p < 0.02), higher mHAI component scores for confluent necrosis and total mHAI score. There were also trends toward a higher percentage of plasma cells in the portal (p < 0.05) and perivenular (p < 0.06) infiltrates, but this did not reach statistical significance. However, the overall severity of portal inflammation, as determined by the mHAI component score, did not differ significantly (p = 0.74).

Table 3. Comparison of statistically significant histological features in native AIH and PCH recipients
 Total PCH groupp-valueIgG4 subgroupsp-value
Native AIH (n = 19)Total PCH (n = 20)IgG4− PCH (n = 11)IgG4+ PCH (n = 9)
  • PCH, plasma cell hepatitis; HAI, histologic activity index; RAI, rejection activity index; NA, not applicable.
  • 1All PCH livers in this category showed 30% plasma cells. However, four of the native liver AIH cases showed <30% periportal plasma cells (20%, 20%, 10% and 10%) and five showed <30% centrilobular plasma cells (20%, 20%, 10%, 10% and 10%).
Banff central perivenulitis severity
None8 (42%)0 (0%)0.0020 (0%)0 (0%)0.0001
Mild7 (39%)6 (30%) 5 (45%)1 (11%) 
Moderate3 (17%)5 (25%) 5 (45%)0 (0%) 
Severe1 (6%)9 (45%) 1 (9%)8 (89%) 
Central fibrosis
NA1 (6%)0 (0%)0.0010 (0%)0 (0%)0.02
None8 (42%)0 (0%) 0 (0%)0 (0%) 
Mild8 (44%)6 (30%) 6 (55%)0 (0%) 
Moderate1 (6%)9 (45%) 4 (36%)5 (56%) 
Severe1 (6%)5 (25%) 1 (9%)4 (36%) 
Centrilobular necrosis
No14 (74%)6 (30%)0.025 (45%)1 (11%)0.16
Yes5 (28%)14 (70%) 6 (55%)8 (89%) 
Overall lobular inflammation severity
None0 (0%)0 (0%)0.090 (0%)0 (0%)0.02
Mild15 (79%)8 (40%) 7 (64%)1 (11%) 
Moderate1 (6%)5 (25%) 3 (27%)2 (22%) 
Severe3 (17%)7 (35%) 1 (9%)6 (67%) 
RAI portal inflammation (0–3)  ± SD0.94 ± 0.81.55 ± 0.80.021.55 ± 0.91.56 ± 0.70.84
RAI bile duct inflammation/damage (0–3) ± SD0.61 ± 0.81.25 ± 0.90.041.27 ± 1.21.22 ± 0.71
RAI venous endothelial inflammation (0–3)  ± SD0.88 ± 0.92.15 ± 1.10.0061.73 ± 1.02.67 ± 10.04
RAI total (#/9) ± SD2.40 ± 2.24.95 ± 2.50.0034.55 ± 2.75.44 ± 2.20.30
HAI confluent necrosis (0–6) ± SD0.37 ± 0.92.3 ± 1.50.0011.82 ± 1.52.89 ± 1.30.08
HAI total (0–18) ± SD7.26 ± 3.19.2 ± 2.80.058.73 ± 3.29.78 ± 2.20.41
Ishak score (0–6) ± SD2.79 ± 1.33.3 ± 1.30.13.09 ± 1.43.78 ± 1.10.14
Periportal plasma cell (%)
≤3012 (63%)5 (25%)10.055 (45%)0 (0%)0.001
31–502 (11%)6 (30%) 5 (45%)1 (11%) 
>505 (28%)9 (45%) 1 (9%)8 (89%) 
Centrilobular plasma cell (%)
NA5 (28%)0 (0%)0.060 (0%)0 (0%)0.0001
≤309 (47%)9 (45%)1 8 (73%)1 (11%) 
31–501 (6%)3 (15%) 3 (27%)0 (0%) 
>504 (22%)8 (40%) 0 (0%)8 (89%) 

Significant differences in the second comparison included the Banff severity of central perivenulitis (p < 0.0001), severity of perivenular fibrosis (p < 0.02), overall lobular inflammation severity (p < 0.02), higher RAI component scores for venous inflammation (p < 0.04) and higher percentage of plasma cells in the portal (p < 0.001) and perivenular (p < 0.0001) infiltrates. Other mHAI component scores, such as interface activity, lobular necro-inflammatory activity and portal inflammation (p = 0.41), did not differ among the groups or between the above comparisons. This shows that overall severity of inflammation does not account for the differences. Characteristic histopathological features of IgG4+ PCH (Figure 2), IgG4− PCH (Figure 3) and native liver AIH (Figure 4) are shown in the figures. All other features examined (Table S1) failed to show significant differences between the groups.

Figure 2.

Example of IgG4+ PCH (Patient 15A). Note the prominent mononuclear portal tract (PT) inflammation, the vast majority of which are mature plasma cells (top frame digital equivalent to approximately ×20; 0.61 microns/pixel2) and inset (digital equivalent to approximately ×60; 0.2 microns/pixel2). Note also that the vast majority of the plasma cells stain positively for IgG4 (bottom frame [digital equivalent to approximately ×20] and inset [digital equivalent to approximately ×60]). There is also very prominent perivenular (CV) plasma-cell-rich inflammation and fibrosis with bridging. *This area is shown at higher magnification in the inset. IgG4—immunohistochemical stain for IgG4-producing plasma cells.

Figure 3.

Example of IgG4− PCH (Patient 12A). Note the prominent portal (PT) and perivenular (CV) plasma-cell-rich inflammation (main frame digital equivalent to approximately ×20; 0.61 microns/pixel2) and inset (digital equivalent to approximately ×60; 0.2 microns/pixel2). In this particular case, perivenular necrosis is also appreciated, but very few IgG4+ plasma cells are detected, despite the prominence of plasma cells in the infiltrate (insets digital equivalent to approximately ×60) *This area is shown at higher magnification in the inset (digital equivalent to approximately ×60). IgG4—immunohistochemical stain for IgG4-producing plasma cells.

Figure 4.

Example of native liver AIH (Patient 6B). Note the portal (PT) and perivenular (CV) plasma-cell-rich inflammation (main frame digital equivalent to approximately ×20; 0.61 microns/pixel2). In this particular case, perivenular plasma-cell-rich inflammation without necrosis is appreciated, but similar to the IgG4− PCH, very few IgG4+ plasma cells are detected (insets digital equivalent to approximately ×60; 0.2 microns/pixel2) *This area is shown at higher magnification in the inset (digital equivalent to approximately ×60). IgG4—immunohistochemical stain for IgG4-producing plasma cells.

Within the IgG4+ PCH recipients, increased numbers of IgG4+ plasma cells and relatively dense lymphoplasmacytic infiltrates and fibrosis were observed, but other histopathological features associated with IgG4-related disease, such as a storiform pattern of fibrosis, definitive obliterative phlebitis and prominent eosinophils, were not detected. The only caveat is whether central perivenulitis and fibrosis might be considered an “obliterative phlebitis,” but certainly not as seen in typical cases of IgG4-related disease.

Evolution of liver injury tests

Liver function tests (LFTs) were performed at index biopsy, 1–2 months following treatment and at last follow-up (Table 4). Interestingly, recipients with IgG4− PCH generally showed higher liver injury tests than those with IgG4+ PCH, but differences were only significant for ALT and AST at the index biopsy (Table 4). More importantly, improvement was seen in both groups after treatment with either corticosteroids and/or increased baseline immunosuppression, as evidenced by the decrease of ALT, AST, ALP, GGTP and T. Bilirubin at all time points afterward (Table 4).

Table 4. Median and mean liver function tests of PCH groups at index biopsy, following treatment and last follow-up
 IgG4− PCH (n = 8–11) IgG4 <25/HPFIgG4+ PCH (n = 7–9) IgG4 ≥25/HPF
Index Bx1–2 moLast FU 8.7–120 moIndex Bx1–2 moLast FU 0.9–40 mo
  • Bx, biopsy; mo, month; FU, follow-up. Comparisons were performed relative to median liver function values within each group (IgG4+ PCH vs. IgG4− PCH, respectively).
  • LFT data from one patient (Patient 3A) in the IgG4− PCH group were removed from the last follow-up time point. This patient was 24 h premortem with markedly elevated ALT (3545 IU/L), AST (4648 IU/L) due to a small bowel infarction, aspiration pneumonia and sepsis, which caused the patient's death.
  • *p ≤ 0.01.
  • **p ≤ 0.05.
ALT U/L median4359236151*4731
Mean (±SD)524 (±502)103 (±89)44 (±26)154 (±104)55 (±29)45 (±30)
AST IU/L median3977133209**4544
Mean (±SD)541 (±592)95 (±90)55 (±44)187 (±94)53 (±30)94 (±119)
ALP IU/L median34916913015816386
Mean (±SD)408 (±295)252 (±177)157 (±59)327 (±258)261 (±237)166 (±197)
GGT IU/L median2521687415514283
Mean (±SD)446 (±344)352 (±316)122 (±166)286 (±358)225 (±176)125 (±110)
TBili mg/dL median1.61.40.951.51.10.9
Mean (±SD)3.7 (±4.6)2.7 (±4.2)6.6 (±12.3)4.2 (±7.7)2.1 (±1.1)6.8 (±16.9)

HCV+ PCH patients

Seven patients with PCH (IgG4+ PCH, n = 4; IgG4− PCH, n = 3) had an original diagnosis of hepatitis C-induced cirrhosis (n = 5) alone, or in association with other diseases (hepatocellular carcinoma, n = 1 and alcoholic liver disease, n = 1). HCV RNA viral load at the time of index biopsy, 1–2 months following treatment and at latest follow-up was correlated to liver function and treatment (immunosuppression and anti-viral treatment) (Figure 5). A majority of these HCV+ patients (57%) were treated by increasing the dose of their immunosuppressant alone or with the addition of another drug added to their standard regime (e.g. MMF or Rapamune [RAPA]). The remaining patients (43%) were treated with corticosteroids. Two patients were exposed to interferon-based therapy prior to their PCH diagnosis (IgG4+ PCH, n = 1; IgG4− PCH, n = 1), similar to previous reports [9]. One of these patients also received therapy with a nucleoside reverse transcriptase inhibitor (Lamivudine [Epivir]) for HBcAb positivity.

Figure 5.

HCV RNA viral load of HCV+ plasma cell hepatitis patients at the time of index biopsy, 1–2 months afterwards, and at last follow-up. Any anti-viral treatment given to each patient is also shown.

Similar to the HCV− PCH patients, HCV+ PCH patients benefited biochemically from treatment with increased baseline immunosuppression with a normalization or improvement of their LFTs regardless of the type of treatment received. HCV RNA viral load remained constant regardless of treatment, except in two patients (corticosteroids [n = 1], increased baseline immunosuppression [n = 1]) who had a 1-log increase in serum HCV RNA levels at 1 month or follow-up. IgG4− PCH patients (n = 3) were treated with steroids, whereas IgG4+ PCH recipients (n = 4) were treated with increased immunosuppression (Figure 5). No patients developed fibrosing cholestatic hepatitis after immunosuppressive treatment.

Association of IgG4+ plasma cells with FOXP3+ cells and CD4+/FOXP3+ cells

Liver biopsy and kidney samples (n = 15; five per cohort) were stained for FOXP3+ cells by single cell immunohistochemistry followed by a more sensitive technique using Qdot CD4+/FOXP3+ double staining (n = 7 livers; three native AIH, four IgG4 PCH). Although there was a trend in the IgG4 PCH livers for increased numbers of FOXP3+ and CD4+/FOXP3+ cells, the cohort was too small to reach statistical conclusions.

Discussion

Liver allograft recipients with PCH are slightly older, show less female predominance and, as a group, show significantly more IgG4-rich infiltrates compared to AIH in native livers or kidney allografts with plasma-cell-rich rejection. Moreover, the trend toward older males in IgG4+ PCH, high prevalence of HLA-DR15 and significant histopathologic differences between IgG4+ PCH and IgG4− PCH on a “blinded” review suggest that IgG4+ PCH likely represents a disorder distinct from native liver AIH and IgG4− PCH. Higher RAI category scores for bile duct damage and central perivenulitis in PCH versus native liver AIH in a “blinded” review also suggest that PCH represents an overlap between allo- and autoimmunity, perhaps related to epitope spreading [26]. Therefore, future studies should search for evidence of cellular- and antibody-mediated pathways of activation against allo-antigens (donor-specific antibodies) and self-antigens (auto-antibodies) to better understand this form of tissue injury.

Polarization of T cell subsets via three different cytokine pathways has the potential to stimulate Breg [10] and other B cells to switch to IgG4-specific plasma cells via production of cytokines and chemokines: (1) Th2 via IL-4 and IL-13 [27]; (2) CD4+/FOXP3+/Treg pathways either via IL-10 and transforming growth factor β (TGFβ) [15] or IL-10 producing Tr1 regulatory cells via GITR/GITR-L, IL-10 and TGFβ [28] and (3) the pro-inflammatory cytokine pathway Th17 via a IL-17, IL-6 and TGFβ [29].

IgG4 is often considered “anti-inflammatory” or “protective” because of its low affinity for C1q and Fc receptors, inability to cross-link antigen abrogating immune complex formation [30] and potential competitive blocking effect on other IgG antibodies [14]. As such, IgG4 usually constitutes only <5% of total serum IgG [31], but can be elevated in response to chronic antigen exposure [32]. Indeed, several studies show that IgG4 can down-modulate the immune system. Extrahepatic cholangiocarcinomas [11] and parasitic infections [12] evade immunosurveillance via the generation of an immunoregulatory cytokine environment rich in IL-10, TGF-β and Treg that promote IgG4 plasma cell production. Attempts at inducing peripheral T cell tolerance in allergic children using allergen-specific immunotherapy follow a similar rationale [13].

The underlying cause of the IgG4-rich plasma cell infiltrates in liver allografts is uncertain, but one can speculate as to several possibilities. Included are: (1) tolerance-promoting unique liver allograft microenvironment or other liver transplant-specific factors such as chronic liver disease, viral infections (CMV, EBV), persistent alloimmune responses and antibody-induction therapies followed by homeostatic proliferation. (2) Increased age, gender and immune senescence-related factors such as shortened telomere length [33-36] and loss of normal cell-mediated and/or humoral homeostatic mechanisms. Indeed, most IgG4+ PCH patients (67%) were males above the age of 50 or post–menopausal women (ages 58–63) who typically show increased IL-6 production, a cytokine hallmark of immune senescence. These patients are prone to more autoimmune events later in life as compared to their younger female counterparts [37, 38]. Advanced age has also been associated with increased serum concentrations of IgG4 [39], and suppressive CD4+/FOXP3+ Tregs have been consistently shown in older T cell populations [40]. (3) Genetic differences might also contribute. A high prevalence of HLA-DR15 was seen in 44% of our PCH cohort overall and in 5/7 patients tested with IgG4+ PCH. HLA-DR15 has also been associated with biliary atresia [41] and primary sclerosing cholangitis [42], which may also be seen in IgG4-related diseases. Additionally, it is also described in multiple sclerosis where increased production of plasma cells and oligoclonal IgG4 bands are seen [43, 44].

The above factors might also act together to promote a “tolerogenic” environment [45, 46], enabling a subset of patients to adapt better to the allograft compared to their counterparts with diminished IgG4, reminiscent of some cases of recurrent HCV where despite heavy infiltration of HCV-specific T cell clones, patients develop operational tolerance [47]. The single patient in this study off immunosuppression is of interest from this perspective.

It should be emphasized that IgG4+ PCH is not synonymous with IgG4-related disease [11, 48, 49]. It is difficult, however, to compare the two groups because of heavy immunosuppression in the former. The IgG4+ PCH patients in this study showed focally dense lymphoplasmacytic infiltrates and nonstoriform fibrosis, but other features of classical IgG4-sclerosing disease [50], such as storiform fibrosis and obliterative phlebitis were not seen and eosinophils were uncommon. We did, however, find an association of IgG4+ PCH with liver fibrosis as in other studies in head, neck and brain, salivary glands, thyroid, lungs, kidneys, retroperitoneum and abdominal organs [23]. Transforming growth factor β (TGFβ), which promotes IgG4 plasma cell development and is a central mediator in fibrogenesis, might be responsible for this observation.

A recent study [51] using next-generation sequencing technology and mutational analysis of the dominant IgG4+ plasma cell clones in paired blood/tissue samples from IgG4-related disease patients showed multiple nonsilent BCR V-region mutations that were different from those seen in healthy and disease control patients. This suggests affinity-driven maturity of the IgG4-producing clones, perhaps directed at self-antigens, which might be applicable to the liver allograft recipients.

Finally, limitations of this manuscript include the size of the cohort, tissue availability for further staining analyses, paucity of routine serological monitoring and the fact that this is a single-center study. Despite these limitations, this study provides a novel observation in PCH allografts that deserves further study to determine the optimal clinical management of these patients and the underlying pathophysiology [52].

Acknowledgment

The authors would like to acknowledge Joe Donaldson, Kim Fuhrer and Lisa Chedwick for their assistance.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary