Plasma cells and the chronic nonsuppurative destructive cholangitis of primary biliary cirrhosis


  • Potential conflict of interest: Nothing to report.

  • Supported by National Institutes of Health grant DK39588.


There has been increased interest in the role of B cells in the pathogenesis of primary biliary cirrhosis (PBC). Although the vast majority of patients with this disease have anti-mitochondrial antibodies, there is no correlation of anti-mitochondrial antibody titer and/or presence with disease severity. Furthermore, in murine models of PBC, it has been suggested that depletion of B cells may exacerbate biliary pathology. To address this issue, we focused on a detailed phenotypic characterization of mononuclear cell infiltrates surrounding the intrahepatic bile ducts of patients with PBC, primary sclerosing cholangitis, autoimmune hepatitis, chronic hepatitis C, and graft-versus-host disease, including CD3, CD4, CD8, CD20, CD38, and immunoglobulin classes, as well as double immunohistochemical staining for CD38 and IgM. Interestingly, CD20 B lymphocytes, which are a precursor of plasma cells, were found in scattered locations or occasionally forming follicle-like aggregations but were not noted at the proximal location of chronic nonsuppurative destructive cholangitis. In contrast, there was a unique and distinct coronal arrangement of CD38 cells around the intrahepatic ducts in PBC but not controls; the majority of such cells were considered plasma cells based on their expression of intracellular immunoglobulins, including IgM and IgG, but not IgA. Patients with PBC who manifest this unique coronal arrangement were those with significantly higher titers of anti-mitochondrial antibodies. Conclusion: These data collectively suggest a role for plasma cells in the specific destruction of intrahepatic bile ducts in PBC and confirm the increasing interest in plasma cells and autoimmunity. (HEPATOLOGY 2012)

Although considerable effort has been expended on defining the pathophysiology of primary biliary cirrhosis (PBC), 1 an interesting void remains, i.e., the relative role of distinct lymphoid populations in chronic nonsuppurative destructive cholangitis (CNSDC) associated with chronic portal inflammation. Indeed, the most disease-specific serologic autoantibodies in all of human immunopathology are the anti-mitochondrial antibodies (AMAs) found in more than 95% of patients with PBC, primarily targeted at the E2 component of the pyruvate dehydrogenase complex (PDC-E2). 2 However, despite the significant value of both AMAs and elevated serum IgM in the diagnosis of PBC, there is no correlation of serum AMA or IgM with either disease severity or any other clinical feature. Furthermore, in murine models of PBC, the clinical use of anti-CD20 antibody, aimed at depleting B cells, has not been successful, and, in another murine model of PBC, depletion of B cells results in escalating liver disease, suggesting that B cells suppress the inflammatory response in mouse models. 3, 4

We have investigated the immunohistochemical distribution of liver-infiltrating B lymphocytes in liver biopsy specimens from patients with PBC and control liver diseases using monoclonal antibody (mAb) reagents specific for CD20 and CD38. We report here a unique coronal arrangement of CD38+ cells that is accompanied by CNSDC. Furthermore, this coronal arrangement of CD38+ cells is positively correlated with AMA titer and is inversely correlated with serum γ-glutamyl transpeptidase (γ-GTP) levels. These CD38+ cells primarily express intracellular IgM or IgG, suggesting a pathogenic role of plasma cells in PBC.


ABC, avidin–biotin–peroxidase complex; AEC, 3-amino-9-ethylcarbazole; AIH, autoimmune hepatitis; ALP, alkaline phosphatase; ALT, alanine aminotransferase; APRIL, a proliferation-inducing ligand; AST, aspartate aminotransferase; AMA, anti-mitochondrial antibody; ANA, anti-nuclear antibody; BAFF, B-cell–activating factor, BD, bile duct; BSA, bovine serum albumin; CH-C, chronic hepatitis C; CNSDC, chronic nonsuppurative destructive cholangitis; CS-1, cell attachment-specific domain 1; EDTA, ethylene diamine tetraacetic acid; ELISA, enzyme-linked immunosorbent assay; γ-GTP, γ-glutamyl transpeptidase; GVHD, graft-versus-host disease; HST, hematopoietic stem cell transplantation; ICAM-1, intercellular adhesion molecule-1; IL, interleukin; IP-10, interferon γ–inducible protein 10; LFA-1, lymphocyte function–associated antigen 1; LIL, liver-infiltrating lymphocyte; mAb, monoclonal antibody; MAdCAM-1, mucosal adressin cell adhesion molecule-1; MIG, monokine induced by interferon γ; NK, natural killer; OLT, orthotopic liver transplantation; PBC, primary biliary cirrhosis; PBL, peripheral blood lymphocyte; PBS, phosphate-buffered saline; PDC-E2, E2 component of pyruvate dehydrogenase; PSC, primary sclerosing cholangitis; SDF-1, stromal cell–derived factor-1; TB, total bilirubin; TBS-T, Tris-buffered saline containing 1% Tween 20; TC, total cholesterol; TG, triglyceride; TP, total protein; UDCA, ursodeoxycholic acid.

Patients and Methods


A total of 78 patients were enrolled in this study. These included 26 patients with PBC, 20 of whom were positive for AMAs and six of whom were negative, 27 patients with chronic hepatitis C (CH-C), eight patients with autoimmune hepatitis (AIH), eight patients with primary sclerosing cholangitis (PSC), and nine patients with graft-versus-host disease (GVHD).

The CH-C control group was age- and sex-matched to the PBC group, which was randomly selected from a cohort of 136 candidates for interferon therapy for CH-C. The diagnosis of all cases was based on established criteria for PBC, 5 AIH, 6 PSC, 7 and GVHD, 8 respectively, or by detection of serum hepatitis C virus RNA by polymerase chain reaction for CH-C. Informed consent in writing was obtained from each patient, and the study protocol was approved by the Institutional Committee for Human Research of Nagaoka Red Cross Hospital. Patient clinical details are presented in Table 1. Liver tissue was available from all patients either from laparoscopic liver biopsies or ultrasound-guided needle liver biopsies.

Table 1. Clinicopathological Profiles of Subjects With PBC, CH-C, AIH, PSC, and GVHD
 PBC n = 26CH-C n = 27AIH n = 8PSC n = 8GVHD n = 9
  • Data are expressed as mean ± standard deviation. For abbreviations, see list in opening page footnote. Statistical significance with Mann-Whitney U test:

  • a

    P = 0.0326 compared with that in PBC; P = 0.0269 compared with that in CH-C; and P = 0.0377 compared with that in AIH.

  • b

    P = 0.0033 compared with that in GVHD.

  • c

    P = 0.0343 compared with that in PBC; P = 0.0139 compared with that in PSC; and P = 0.0003 compared with that in GVHD

  • d

    P = 0.0119 compared with that in GVHD.

  • e

    P = 0.0038 compared with that in GVHD.

  • f

    P = 0.0019 compared with that in GVHD.

  • g

    P = 0.0233 compared with that in GVHD.

  • h

    P = 0.0035 compared with that in PBC; and P = 0.0004 compared with that in CH-C.

  • i

    P = 0.0015 compared with that in PBC; and P = 0.0001 compared with that in CH-C and P = 0.0460 compared with that in PSC.

  • j

    P = 0.0148 compared with that in CH-C.

  • k

    P = 0.0184 compared with that in CH-C.

  • l

    P < 0.0001 compared with that in CH-C; P = 0.0133 compared with that in AIH; and P = 0.0082 compared with that in GVHD.

  • m

    P = 0.0085 compared with that in CH-C.

  • n

    P < 0.0001 compared with that in CH-C; and P = 0.0008 compared with that in AIH.

  • o

    P = 0.0002 compared with that in CH-C.

  • p

    P = 0.0004 compared with that in CH-C.

  • q

    P < 0.0001 compared with that in CH-C; and P = 0.0433 compared with that in AIH.

Age, years57.2 ± 9.357.9 ± 11.359.5 ± 9.546.0 ± 24.645.9 ± 14.2a
Scheuer stage, 1/2/3/415/3/8/0
Fibrosis score, 0/1/2/3/40/15/6/6/00/2/5/1/01/7/0/1/0
Activity score, 0/1/2/30/13/12/20/1/1/60/8/0/1
Ludwig stage, 1/2/3/45/1/1/1
Biopsy procedures, Echo/Laparo/Ope1/24/127/0/01/7/06/2/09/0/0
Evaluated portal tracts, no.12.2 ± 4.7b15.2 ± 5.6c13.0 ± 4.2d9.0 ± 6.57.0 ± 2.5
Evaluated bile ducts, no.22.5 ± 11.4e25.6 ± 11.6f30.6 ± 19.8g17.5 ± 15.112.0 ± 4.1
AST, IU/L99.3 ± 141.155.7 ± 40.9332.1 ± 522.2h111.9 ± 90.7124.4 ± 140.8
ALT, IU/L105.2 ± 159.458.7 ± 42.4333.4 ± 367.2i126.4 ± 99.1j217.4 ± 289.0k
ALP, IU/L775.7 ± 389.4l251.2 ± 92.3412.0 ± 156.9m560.8 ± 627.2434.2 ± 348.4
γ-GTP, IU/L380.1 ± 258.8n48.7 ± 43.4155.1 ± 36.4o374.3 ± 269.8p572.3 ± 668.7q

Liver Histology and Clinical Data Collection.

Liver biopsy specimens were fixed in 10% formalin, dehydrated, and embedded in paraffin. Then 5-μm sections were cut in a microtome and subjected to subsequent routine histological staining using silver impregnation, hematoxylin and eosin, and diastase-resistant periodic acid Schiff. Histological staging based on established criteria for PBC, 5 CH-C, 9 AIH, and PSC 10 was performed by a pathologist who was blinded to all clinical data. Blood biochemical data were obtained from each patient within 1 week of liver biopsy. These included aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), and γ-GTP. Total bilirubin, indirect bilirubin, total protein, serum albumin, total cholesterol (TC), triglycerides, serum levels of IgG, IgA, and IgM, AMA titers, and anti-nuclear antibody titers were included in the PBC cohort. AMAs were examined by immunofluorescence and by titration with enzyme-linked immunosorbent assay (ELISA) with known positive and negative standards throughout. Anti-nuclear antibodies were determined by immunofluorescence on Hep2 cells.


Immunohistochemistry of liver biopsy was performed with a Ventana HX System BenchMark/20 (Ventana Medical Systems Inc., Tucson, AZ), which uses avidin-biotin-peroxidase complex (ABC) coupled with a formulated unmasking pretreatment for each targeted antigen. For unmasking CD antigens and pankeratins, liver sections were soaked in Tris-EDTA buffer, pH 8.5 (Ventana CC1 standard solution) at 100°C for 60 minutes. For unmasking immunoglobulins, liver sections were soaked in Protease 1 solution (pronase 0.5 U/mL) for 8 minutes. Endogenous peroxidase activity in liver tissue was blocked by soaking specimens in 3% H2O2 methanol solution for 10 minutes. Nonimmunized mouse IgG, used as a negative control in each experiment, did not result in any nonspecific staining signal.

The following mouse monoclonal antibodies were used: anti-CD3 (clone PS1), anti-CD4 (clone 1F6), anti-CD8 (clone 1A5), and anti-CD38 (clone SPC32) from Novacastla Laboratories (Newcastle upon Tyne, UK); anti-CD20 (clone L26) and anti-pankeratin (clone AE1/AE3) from Dako (Glostrup, Denmark); and anti-human IgG (clone A57H), anti-human IgA (clone CB1-10.4/B8), and anti-human IgM (clone R1/69) from Nichirei Corp. (Tokyo, Japan). After incubation with each of these primary antibodies at an appropriate dilution with 5% bovine serum albumin (BSA), slides were rinsed three times with phosphate-buffered saline (PBS), and then incubated with biotinylated anti-mouse IgG secondary antibodies. The slides were rinsed three times and then incubated with ABC reagents for staining of CD antigens and pankeratin. For staining of immunoglobulins, we used standard peroxidase-labeled streptavidin-biotin. Diaminobenzidine hydrochloride was used as a substrate for colorimetric reaction.

Portal tracts and bile ducts were counted per specimen with pankeratin staining using AE1/AE3 monoclonal antibodies that specifically stain bile ducts and bile ductules. Only portal tracts with more than half of its circumference and proper bile ducts were counted, and ductular reactions were excluded from this count. The bile ducts with a coronal arrangement of CD38+ cells were enumerated per specimen, and the frequency of this finding among the total counted bile ducts was thence calculated.

For CD38 and IgM double immunohistochemical staining, mouse monoclonal anti-CD38 (SPC32) and rabbit polyclonal anti-IgM (Pierce Biotechnology, Rockford, IL) were used as previously described. 11 After deparaffinization, sections were soaked in target retrieval–buffered saline (Tris, pH 6.1, Dako Cytomation, Carpinteria, CA) in a plastic pressure cooker containing no metals, irradiated in a microwave oven for 10 minutes, soaked in 3% H2O2 methanol solution for 5 minutes, and then soaked in 5% BSA for 1 minute. A cocktail of anti-CD38 and anti-IgM antibodies was diluted to a predetermined optimal concentration in PBS containing 5% BSA. The diluted antibodies were applied to tissue sections in a moist chamber and irradiated intermittently for 10 minutes (250 W, 4 seconds on, 3 seconds off). After three washes with Tris-buffered saline containing 1% Tween 20 (TBS-T) for 1 minute, a cocktail containing peroxidase-conjugated (Envision System, Dako Cytomation) or alkaline phosphatase–conjugated secondary antibodies (Simple Stain System, Nichirei, Japan) was applied to specimens in the moist chamber. Irradiation was then performed intermittently for 10 minutes, as described above. After washing five times with TBS-T, the sections were immersed in Fast blue (alkaline phosphatase substrate kit, SK-5300, Vector, Burlingame, CA) and 3-amino-9-ethylcarbazole (AEC; peroxidase substrate kit, Nichirei) and counterstained with hematoxylin (Dako Cytomation). After the substrate reaction, CD38+ cells were blue and IgM+ cells were red–brown.


The Mann-Whitney U test was used for comparing the blood biochemical and serological data among PBC, CH-C, AIH, PSC, and GVHD groups and between PBC patients with or without a coronal arrangement of CD38+ cells; P values under 0.05 were considered statistically significant. The Chi-square test was used for comparing the occurrence of lymph follicle–like infiltration and the infiltration of CD4+ or CD8+ cells into cholangioepithelium among the PBC, CH-C, AIH, PSC, and GVHD groups.


Clinicopathological profiles of PBC, CH-C, AIH, PSC, and GVHD subjects are summarized in Table 1.

Therapy for the 26 PBC patients varied according to the nature and severity of the disease. Ursodeoxycholic acid (UDCA) was administered to 25 patients (96.2%); the daily dose was 300 mg in one, 600 mg in 22, and 900 mg in two patients. Bezafibrate was also administered to four patients at a daily dose of 400 mg, all of whom simultaneously took UDCA. Prednisolone was given to two patients in whom a hepatitic form of PBC was noted on liver pathology. None of the 26 PBC patients progressed to the icteric stage, which would need liver transplantation during the observation period through September 2011.

Histologic evaluations of liver biopsy sections revealed typical CNSDC and epithelioid cell granulomas in patients with PBC but not those with the control liver diseases (Table 2). The incidence of CNSDC in PBC was 50% (13/26) and that of granulomas was 23.1% (6/26). In contrast, lymph follicle–like infiltration was most frequently found in CH-C livers (12/27, 44.4%) but was also found at lower frequencies in patients with PBC (6/26, 23.1%) and AIH (2/8, 25%). Such infiltration was not found in PSC or GVHD.

Table 2. Histological and Immunohistological Characteristics of the PBC, CH-C, AIH, PSC, and GVHD Livers
 PBC n = 26CH-C n = 27AIH n = 8PSC n = 8GVHD n = 9
 No. of cases%No. of cases%No. of cases%No. of cases%No. of cases%
  • For abbreviations, see list in opening page footnote.

  • *

    Statistical significance with Chi-square tests: P = 0.0221 compared with PBC; P = 0.0017 compared with that in CH-C; and P = 0.0275 compared with that in PSC.

Lymph follicle–like infiltration*6(23.1)12(44.4)2(25.0)0000
Infiltration of CD4+ cells into cholangioepithelium*16(61.5)001(12.5)2(25.0)00
Infiltration of CD8+ cells into cholangioepithelium*7(26.9)4(14.8)3(37.5)1(12.5)7(77.8*)
Coronal arrangement of CD38+ cells13(50.0)0000001(11.1)
Coronal arrangement of IgG+ cells5(19.2)
Coronal arrangement of IgM+ cells8(30.8)
Coronal arrangement of IgG+ and IgM+ cells3(11.5)

The distribution of lymphoid elements is summarized in Table 2. In PBC, CD20+ B lymphocytes, the precursors of plasma cells, were found either scattered or aggregated within the lymphoplasmocytic infiltration (Fig. 1A). Such CD20+ B cells occasionally formed follicle-like aggregations but, importantly, they were not observed in the proximity of CNSDC (Fig. 2B). In contrast, an intense coronal arrangement of CD38+ cells was found around intrahepatic bile ducts in every specimen with CNSDC (Figs. 1B, 2C) but was never found in the portal tracts with ductopenia. CD3+ pan-T cells were randomly scattered around CNSDC (Fig. 2A), an area where CD20+ B lymphocytes had not been observed (Fig. 2B), whereas the most typical coronal arrangement of CD38+ cells was found (Fig. 2C). This coronal arrangement of CD38+ cells was continuously present along with CNSDC, as shown by serial liver sections (Fig. 2D). CD4+ and CD8+ T lymphocyte infiltration was observed either in proximity to, or within, the degenerated cholangioepithelium, suggesting the participation of these cells in the destructive processes of intrahepatic bile ducts (Fig. 2E,F). This CD4+ and CD8+ T lymphocyte infiltration into the cholangioepithelium was also observed in patients with CH-C, AIH, PSC, and GVHD as a consequence of lymphocytic cholangitis (Table 2).

Figure 1.

Immunohistochemical staining of CD20+ B lymphocytes and CD38+ cells in consecutive sections of a PBC liver. (A) CD20+ B lymphocytes are either aggregated in lymph follicle–like structures (white stars) or scattered around inflamed portal tracts. (B) Coronal arrangement (CA) of CD38+ cells surrounding the intralobular bile ducts (arrows). Note that CD38+ cells are scarce in lymph follicle–like structures. ABC method, original magnification, x10.

Figure 2.

Immunohistochemical staining of CD3+ pan-T lymphocytes, CD20+ B lymphocytes, CD38+ cells, CD4+ T lymphocytes, and CD8+ T lymphocytes in two sets of consecutive PBC liver sections. (A-D) In the first consecutive section set, CD3+ pan-T lymphocytes are scattered around an intrahepatic bile duct (BD) with CNSDC (A); CD20+ B lymphocytes are not observed around this intrahepatic bile duct (BD; B); and CD38+ cells show a distinct coronal arrangement surrounding this intrahepatic bile duct (BD) with CNSDC (C). The coronal arrangement of CD38+ cells is continuously observed along with CNSDC in serial sections (D). In subsequent serial consecutive sections, CD4+ T lymphocytes (E) and CD8+ T lymphocytes (F) demonstrate the same pattern of infiltration (arrows) into the cholangioepithelium of an intrahepatic bile duct (BD) with CNSDC. ABC method, original magnification ×25.

To determine whether the coronal arrangement pattern of CD38+ cells was specific for PBC, we examined CD20+ and CD38+ cells in liver sections with other liver diseases. In CH-C, CD20+ B lymphocytes were aggregated in a follicle-like fashion in the inflamed portal tracts (Fig. 3A) where intrahepatic bile ducts were often centered (an arrow in Fig. 3A); in contrast, CD38+ cells were found at the periphery of inflamed portal tracts but were not found around the intrahepatic bile ducts (arrow in Fig. 3B). Similarly, in AIH, CD38+ cells were not observed in the proximity of intrahepatic bile ducts (BD, arrows in Fig. 3C) but were abundantly infiltrated in the area of interface hepatitis (Fig. 3C). In contrast to PBC livers, in which CD38+ cells formed a coronal arrangement around an intrahepatic bile duct (BD) with CNSDC (Fig. 3D), such a pattern was not observed in the disease control groups including CH-C, AIH, and PSC; it was observed in only one bile duct of one patient with GVHD (Table 2), although the frequency of this finding was only 0.9% (1/107) of all bile ducts evaluated in this group and was thought to be incidental. In PBC livers, the coronal arrangement of CD38+ cells was observed in 21.5% (7.1%-41.0% per specimen) of all evaluated bile ducts when it was present (69 bile ducts among 321 bile ducts counted in the PBC group with a coronal arrangement of CD38+ cells). In PSC livers, CD38+ cells were found surrounding, or in an onion skin–like fibrosis, a pattern that is unique in PSC (Fig. 3E,F).

Figure 3.

Immunohistochemical staining of CD20+ and CD38+ cells in liver sections. (A) In CH-C, CD20+ B lymphocytes are aggregated in a follicle-like structure in an inflamed portal tract (white stars). Note that an intrahepatic bile duct (arrow) is centered in such a follicle-like structure. (B) In a consecutive section of CH-C liver (with A), CD38+ cells are primarily located in the periphery of inflamed portal tracts (black stars) apart from the intrahepatic bile ducts (an arrow). (C) In AIH, CD38+ cells are not observed in the proximity of intrahepatic bile ducts (BD, arrows) but are abundantly infiltrated in the area of interface hepatitis (black stars). (D) In PBC, CD38+ cells show the coronal arrangement surrounding intrahepatic bile ducts (BD) with CNSDC (arrows). (E) In a case of stage 4 PSC, CD38+ cells are surrounding the onion skin-like fibrosis that is a characteristic of PSC (arrows). This pattern is termed a satellite-like arrangement (SA) of CD38+ cells surrounding concentric periductal fibrosis. The SA pattern is different from the CA of CD38+ cells in PBC located just beneath the cholangioepithelium with CNSDC, as shown in Figs. 2C and 4D. (F) In a case of stage 1 PSC, CD38+ cells are concentrically scattered in the onion skin–like fibrosis (black stars) surrounding an intrahepatic bile duct (arrows). This pattern is also regarded as SA of CD38+ cells. ABC method, original magnification ×25.

To determine the identity of the CD38+ cells that formed a coronal arrangement specifically in the PBC livers with CNSDC (Fig. 2C), we first examined the expression of immunoglobulin classes in consecutive sections of PBC livers with CNSDC by staining for immunoglobulin classes. The majority (8/13, 61.5%) of the coronal arrangements had IgM+ cells (Fig. 4C), followed by IgG+ cells (5/13, 38.5%; Fig. 4A). Three patients (23.1%) had both IgG+ and IgM+ cells in the same coronal arrangement (Fig. 4) whereas two had only IgG+ cells but not IgM+ cells. IgA+ cells were not observed in a coronal arrangement (Fig. 4B). IgG+ cells were relatively loosely scattered around CNSDC (Fig. 4A), whereas IgM+ cells showed a dense and prominent coronal arrangement around CNSDC (Fig. 4C). These results indicate that the B cells expressing antibodies participate in the formation of the coronal arrangement in CNSDC.

Figure 4.

Immunohistochemical staining of IgG, IgA, and IgM in consecutive liver sections of a PBC liver. (A) IgG+ cells demonstrate a relatively loose coronal arrangement surrounding an intrahepatic bile duct (BD) with CNSDC. (B) IgA+ cells are not found in the vicinity of this intrahepatic bile duct (BD) with CNSDC. (C) IgM+ cells demonstrate a dense and distinct coronal arrangement surrounding this intrahepatic bile duct (BD) with CNSDC. Labeled streptavidin-biotin method, original magnification ×25.

Next we used double immunostaining to examine the colocalization of CD38 and IgM in the same cells that comprise the coronal arrangement. The majority (up to 70%) of CD38+ cells in coronal arrangement that were specifically stained by Fast blue also showed positive red–brown staining of intracellular IgM (arrows in Fig. 5). These results suggest that the majority of CD38+ cells were IgM plasma cells. Finally, we examined the correlation of coronal arrangement with blood biochemical and serological parameters in PBC (Table 3). Among 14 parameters, the presence of coronal arrangement was significantly associated with higher titers of AMA (P = 0.0153) and lower levels of γ-GTP (P = 0.0256) (Fig. 6).

Figure 5.

Double immunostaining of CD38 and IgM in a PBC liver. The majority (approximately 70%) of CD38+ cells that are immunohistochemically identified by a blue color (Fast blue) demonstrate a reddish brown color (AEC) in their cytoplasm indicating the presence of intracellular IgM (arrows). BD, an intrahepatic bile duct with CNSDC, indirect double enzyme-antibody method, original magnification ×100.

Figure 6.

AMA titers and γ-GTP levels in coronal arrangement (CA)-positive and CA-negative PBC patients . (A) The titer of AMA is significantly higher in the CA-positive PBC patients compared with the CA-negative patients. (B) γ-GTP level is significantly lower in the CA-positive PBC patients compared with the CA-negative patients.

Table 3. Coronal Arrangement (CA) of CD38+ Cells and Clinical Data in PBC
 CA (−) n = 13CA (+) n = 13P
  • CA (−), without coronal arrangement of CD38+ cells; CA (+), with coronal arrangement of CD38+ cells. For other abbreviations, see list in opening page footnote.

  • *

    P < 0.05 (significant) by the Mann-Whitney U test.

AST88.6 ± 91.4110.0 ± 181.30.8174
ALT88.6 ± 79.1121.8 ± 215.50.7778
ALP789.8 ± 422.4761.7 ± 370.31.0000
γ-GTP477.6 ± 296.4282.5 ± 176.20.0256*
TB0.63 ± 0.220.71 ± 0.180.2012
TP7.67 ± 0.58.02 ± 0.770.1635
Albumin4.22 ± 0.314.35 ± 0.310.2668
TC223.6 ± 34.7195.1 ± 32.40.1059
TG116.2 ± 56.8111.5 ± 50.10.9795
IgG1,728.3 ± 367.21,838.2 ± 529.40.7779
IgA266.4 ± 79.4298.5 ± 112.30.3428
IgM293.5 ± 199.9517.2 ± 438.30.0858
AMA titer67.9 ± 61.8119.5 ± 58.20.0153*
ANA titer433.8 ± 791.3163.1 ± 3550.7163


The study of plasma cells in autoimmune diseases has led to the hypothesis that plasma cells with pathogenic potential are long-lived and depend on finding a niche within a local microenvironment such as the biliary tract. Indeed, in murine lupus, B-cell–activating factor (BAFF), a proliferation-inducing ligand (APRIL), interleukin-6 (IL-6), and adhesion molecules all modulate the survival of plasma cells and lead to further inflammation. 12-16 In organ-specific autoimmune diseases such as PBC, the role of plasma cells has not attracted significant attention. Here we have demonstrated a relatively nonspecific follicle-like aggregation in inflamed portal tracts of CD20+ B cells and, more importantly, a prominent coronal arrangement of CD38+ plasma cells surrounding the intrahepatic bile ducts with CNSDC. CD20 is a type III membranous protein of 297 amino acids 17; it is a representative B-lineage cell marker that disappears from the cell surface when B cells are differentiated into antibody-producing plasma cells. Therefore, the differential distribution of CD20+ and CD38+ cells in a target tissue reflects the disease-specific movement of these two cell types during B-cell maturation in the course of a chronically evolving inflammatory disease such as PBC.

Our findings suggest a PBC-specific dynamic settlement in B-cell lineage populations during the inflammatory processes in portal tracts, which involves migration of B lymphocytes from the portal tracts to the intrahepatic bile ducts during the maturation process from CD20+ mature B cells to professional antibody-producing CD38+ cells, or plasma cells. We did not, however, examine CD38+ cells in cirrhotic livers of patients with PBC. Future studies should focus on a longitudinal analysis and/or detailed cross-sectional analysis of patients at different stages of disease. Also, this concept of plasma cells infiltrating environmental niches needs further exploration in other liver diseases, such as expanding the PSC database and including patients with hepatic allograft rejection after orthotopic liver transplantation (OLT) and chronic GVHD after hematopoietic stem cell transplantation (HST).

The most important finding of this study is the coronal arrangement pattern of CD38+ cells around the intrahepatic bile ducts with CNSDC. This pattern is specific for PBC and is not observed in other autoimmune liver diseases including AIH, PSC, and GVHD, or in CH-C. CD38 is a part of nicotinamide adenine dinucleotide cyclase. 18 It is a type II membranous protein of 300 amino acids. CD38 expression is limited to the cell surface of T-cell progenitors, lymphoid stem cells, plasmablasts, and mature plasma cells. We conclude that the majority of CD38+ cells observed around the intrahepatic bile ducts are mature plasma cells rather than T cell subsets based on the following observations: 1) the distributions of CD38+ cells and of IgM- and/or IgG-bearing cells were nearly identical; 2) the distributions of CD3+, CD4+, and CD8+ T cells and that of CD38+ cells were very different; 3) mature plasma cells expressed higher levels of CD38 than other CD38+ B-cell or T-cell subsets 19, 20; 4) the distribution of CD138+ cells, a marker more specific for mature plasma cells than CD38, was similar to that of CD38+ cells (data not shown); and 5) most importantly, double immunostaining of CD38 and IgM in PBC livers clearly indicated that approximately 70% of CD38+ cells expressed IgM. However, there remains a possibility that the observed CD38+ population was a mixture of diverse cell types including activated T cells, other B-lineage cell populations, natural killer (NK) cells, and basophils in addition to mature plasma cells. 21 Even if this is the case, the fact that CD38+ cells clearly form a coronal arrangement surrounding CNSDC strongly suggests that the coronal arrangement is related to the pathogenesis of PBC.

The role of T-cell lineage populations in the pathogenesis of PBC has been extensively studied. 1, 22-31 In contrast, the role of B-cell lineage populations in PBC is not clear. Recently it has been shown that AMAs are required in the production of inflammation cytokines by macrophages in the presence of apoptotic human intrahepatic biliary epithelial cells. 32 This could explain the biliary specificity of autoimmune damage in PBC. It is possible that at different stages of PBC, B cells play different roles in the breakdown of tolerance and development of small bile duct pathogenesis. Although we cannot exclude the possibility that the formation of coronal arrangement is a consequence of, rather than a contributing factor to, the destruction of small bile ducts, our data are in agreement with the findings of Lleo et al., 32 which strongly suggest an active role of AMAs in the inflammatory responses at the affected local bile ducts. Future studies should focus on the antigen specificity of the IgM and IgG produced in these coronal arrangement–comprising plasma cells. Although IgA transcytosis has been considered one of the contributing factors toward bile duct lesions in PBC, 33, 34 it is unlikely that the coronal arrangement we observed in this study includes IgA+ plasma cells. However, we observed that IgA staining was often seen in the degenerated cholangioepithelium of CNSDC or at the apical margin of damaged bile duct cells (data not shown).

There are two steps in lymphocyte recruitment and chemotaxis in PBC: “tethering” and transendothelial migration of peripheral blood lymphocytes (PBLs) from vessels to the inflamed area of liver, and recruitment and settlement of liver-infiltrating lymphocytes (LILs) around bile ducts, where they participate in the destruction process of targeted bile ducts. Induced or up-regulated expression of monokine induced by interferon γ (MIG) and interferon γ–inducible protein 10 (IP-10) in portal tracts is thought to contribute to T-cell recruitment into the PBC liver. 35 Once these cells have entered the portal tract, they often form lymphoplasmocytoid aggregates. Lymphoid neogenesis depends on the interaction between CCL21 36 and mucosal adressin cell adhesion molecule-1 (MAdCAM-1) 37 expressed on high endothelial venules and on lymphatic vessels and CCR7 expressed on lymphocytes. 38 The lymphoplasmocytoids are then recruited to, and retained around, bile ducts by the combinational or sequential action of several chemokines. 35, 39, 40 In the case of the B cells, only CXCL12 (stromal cell-derived factor 1 [SDF-1]) has been reported to participate in the recruitment of CD19+ B cells 41-43; however, CD19 is expressed in intermediate and mature B cells but not in plasma cells.

CXCL12 (SDF-1) is expressed constitutively in bile duct cells and may have a role in retention of lymphocytes via its ability to augment their adhesion to fibronectin by triggering α4β7-mediated binding of LIL. 38, 44, 45 Moreover, the biliary basement membrane has immunoreactive fibronectin in 80% of patients with PBC but not in those with other diseases or in normal control livers. 46 We previously found that fibronectin was abundantly expressed in necrotic and newly fibrosing areas, or in the area of inflammation, 47 suggesting that fibronectin may contribute to the chemotaxis of CD38+ cells in interface hepatitis in CH-C and AIH liver as well as in coronal arrangement in PBC livers. In addition, we have shown that the expression of alternatively spliced fibronectin containing a cell attachment–specific domain (CS-1 fibronectin) was increased in fibrotic human liver 48; hence it may also contribute to the recruitment and retention of LIL through not only the α1β4 but also the α4β7 integrin. Taken together, these previous findings suggest that CXCL12 might be involved in the formation of the coronal arrangement of CD38+ cells around bile ducts. Other potential contributing factors for the formation of coronal arrangement include hepatocyte growth factor, 49 lymphocyte function–associated antigen-1 (LFA-1)/ intercellular adhesion molecule-1 (ICAM-1), 50, 51 CXCL13, and CXCL5, 52 which should also be explored in future studies.

Although AMAs are well established as a diagnostic criterion for PBC, serum AMA titer has not been correlated with any parameter of disease activity. We observe for the first time that AMA titer was correlated significantly with the formation of a coronal arrangement by CD38+ plasma cells in the liver, which could be a bridge that links AMA and the inflammatory reactions in PBC. Although the implications of this novel correlation need to be studied further, we propose that plasma cells may be a source of both AMAs and elevated serum IgM. UDCA is reported to decrease the level of total IgM and IgM AMAs but not of IgG AMAs. 53 Thus, UDCA may preferably affect the CD38+ plasma cells that comprise the coronal arrangement. The relationship of reduced γ-GTP with the coronal arrangement is unclear because another biliary enzyme, ALP, did not correlate with the coronal arrangement. Finally, the concept of an environmental niche for long-lived plasma cells and their impact on inflammation has implications not only for PBC, but, from a generic perspective, for other autoimmune diseases as well; this concept has been recently reviewed. 54


The authors thank Mr. Kyuji Iwamoto for his technical assistance in preparing histological sections and immunohistochemistry.