Isolated Donor Specific Alloantibody-Mediated Rejection after ABO Compatible Liver Transplantation


*Corresponding author: Kenneth A. Andreoni,


Antibody-mediated rejection (AMR) after liver transplantation is recognized in ABO incompatible and xeno-transplantation, but its role after ABO compatible liver transplantation is controversial.

We report a case of ABO compatible liver transplantation that demonstrated clinical, serological and histological signs of AMR without evidence of concurrent acute cellular rejection. AMR with persistently high titers of circulating donor specific antibodies resulted in graft injury with initial centrilobular hepatocyte necrosis, fibroedematous portal expansion mimicking biliary tract outflow obstruction, ultimately resulting in extensive bridging fibrosis. Immunofluorescence microscopy demonstrated persistent, diffuse linear C4d deposits along sinusoids and central veins. Despite intense therapeutic intervention including plasmapheresis, IVIG and rituximab, AMR led to graft failure.

We present evidence that an antibody-mediated alloresponse to an ABO compatible liver graft can cause significant graft injury independent of acute cellular rejection. AMR shows distinct histologic changes including a characteristic staining profile for C4d.


Convincing evidence of negative graft outcomes has been reported after ABO incompatible liver transplantation (1). In this ABO incompatible population, initial reports of the histological features of the humoral component were indistinguishable from severe ischemia reperfusion injury (2). More recent reports note additional distinctive histological features such as periportal edema and necrosis (3). The antibody response to ABO incompatibility is based on isohemaglutinins to carbohydrate moieties and is not a response to HLA antigens. Despite this difference in alloantibodies versus isohemaglutinins, there is histological similarity between ABO incompatible and compatible humoral injury in renal transplantation (4).

In contrast, antibody-mediated rejection (AMR) does not appear to play a significant role after ABO compatible liver transplantation. Traditionally, the liver is considered an “immune privileged organ”, resisting humoral rejection caused by preformed HLA antibodies (5). Numerous studies show conflicting graft outcomes using ABO compatible livers in the setting of a positive HLA crossmatch (Table 1). This disparity in observations may be related to different response to antibody-mediated injury caused by diverse antibody affinity, specificity and abundance of target donor antigens on the graft endothelium. A few authors have reported AMR after liver transplantation using ABO compatible donors (6,7). None of the reports, however, fulfills all criteria of AMR put forward to standardize diagnoses in renal and cardiac transplantation by a multidisciplinary group. Criteria for liver allograft AMR was not defined (8).

Table 1.  Publications presenting differences in long-term graft loss after ABO compatible liver transplantation with a positive crossmatch
Increased graft lossNo increased graft loss
1. Bathgate AJ, et al. Liver Transpl Surg 4: 280–284, 1998 1. Dagher L, et al. Transplant Proc 31: 2363, 1999
2. Bishara A, et al. Hum Immunol 63: 742–750, 2002 2. Donaldson PT, et al. Transplantation 60: 1016-1023, 1995
3. Charco R, et al. Transpl Int 7 Suppl 1: S94–S96, 1994 3. Doyle HR, et al. Ann Surg 224: 168–177, 1996
4. Charco R, et al. Transplant Proc 29: 461–462, 1997 4. Fujita S, et al. Transplantation 63: 84–88, 1997
5. Charco R, et al. J Hepatol 24: 452–459, 1996 5. Goggins WC, et al. Transplantation 62: 1794–1798, 1996
6. Dawson S, 3rd, et al. Clin Transpl: 18: 189–195, 1994 6. Gordon RD, et al. Gastroenterol Clin North Am 17: 53–59, 1988
7. Dawson S, 3rd, et al. Artif Organs 20: 1063–1072, 1996 7. Gordon RD, et al. Surgery 100: 705–715, 1986
8. Demetris AJ, et al. Hepatology 16: 671–681, 1992 8. Lang M, et al. Transplant Proc 30: 1466–1467, 1998
9. Doran TJ, et al. Transplantation 69: 1491–1498, 2000 9. Matinlauri IH, et al. Clin Transplant 19: 626–631, 2005
10. Evrard V, et al. Ann Surg 239: 272–280, 200410. Starzl TE, et al. Transplant Proc 6: 129–139, 1974
11. Gugenheim J, et al. Transplant Proc 31: 552–553, 199911. Starzl TE, et al. Gastroenterology 77: 375–388, 1979
12. Hathaway M, et al. Transplantation 64: 54–59, 1997 
13. Karuppan S, ET AL. Transpl Int 4: 18–25, 1991 
14. Katz SM, et al. Transplantation. 57: 616–620, 1994 
15. Muro M, et al. Transpl Immunol 14: 91–97, 2005 
16. Muro M, et al. Transplant Proc 35: 1894–1895, 2003 
17. Nikaein A, et al. Transplantation 58: 786–792, 1994 
18. Ogura K, et al. Clin Transplant 8: 111–115, 1994 
19. Takaya S, et al. Transplantation 53: 400–406, 1992 
20. Takaya S, et al. Transplant Proc 31: 2028–2031, 1999 

We report the first documented case of AMR without acute cellular rejection that fulfills all required criteria of humoral alloinjury in an ABO compatible setting as proposed in renal and cardiac transplantation (8). We define distinct histologic changes and C4d staining patterns that closely resemble observations made in renal allografts (9,10). Despite attempts to remove circulating donor specific antibodies (DSA) in the case under discussion, AMR ultimately resulted in graft fibrosis and loss.

Materials and Methods


Liver tissue was fixed in 10% buffered formalin. Sections 5μm thick were cut from paraffin-embedded tissue cores, and stained with hematoxylin and eosin (H&E) and trichrome stains for morphologic examination. Immunofluorescence studies were performed on liver tissue rapidly frozen in precooled isopentane and cut in 4 μm thick sections. C4d was detected using a mouse monoclonal antibody (Quidel, San Diego, CA) according to a previously published protocol (10). C4d staining was performed on fresh frozen biopsies obtained at POD 0 after liver reperfusion and POD 40.

HLA typing

The patient's HLA-A, B, and DR type was determined by PCR-SSP with commercially available reagents (Dynal, Inc., Brown Deer, WI). The deceased donor HLA type was confirmed by serologic-based typing with commercially available Class I and II HLA typing trays (One Lambda, Inc., Canoga Park, CA).

Antibody screening and crossmatching

Allosensitization to HLA antigens was determined by anti-human globulin (AHG) augmented complement-dependent cytotoxicity (CDC) using commercially available T-cell trays (Gen-Trak, Inc., Liberty, NC). Alloantibody profiles were also assessed by flow cytometry with FlowPRA Class I and II single antigen beads (One Lambda, Inc., Canoga Park, CA).

CDC crossmatches with magnetic bead purified donor B cells were performed using an extended incubation method and with magnetic bead purified donor T cells using Amos Modified and AHG-augmented assays. Flow cytometric crossmatches against pronase treated donor lymphocytes were performed using a three-color technique (anti-IgG F (ab`)2 FITC, CD19PE and CD3PerCP) using a FACS can flow cytometer (BD Biosciences, Inc., Rockville, MD). Patient sera were incubated with anti-rabbit agarose beads prior to crossmatching to remove thymoglobulin. Pooled Sera from patients with a high PRA were used as a positive control and serum from a nontransfused healthy male screened for the presence of HLA alloantibody was used as a negative control.

Case Report

The patient was a 50-year-old Caucasian female, blood type O negative, with alcoholic cirrhosis and hepatopulmonary syndrome.

She underwent an ABO compatible, five antigen mismatch, deceased donor liver transplantation from a 47-year-old female, O positive, with negative hepatitis serologies. Induction therapy consisted of thymoglobulin on postoperative days (POD) 0, 3, 4 and 5 to delay calcineurin inhibitor therapy due to preoperative renal dysfunction. A postreperfusion 1 h biopsy was obtained that showed mild confluent centrilobular necrosis (Figure 1A). C4d was detected with a diffuse strong linear staining pattern along sinusoidal spaces and endothelial surfaces of central veins (Figure 3A). The T-cell AHG-CDC crossmatch from preoperative serum was positive at 1:1 dilution, with a flow T-cell channel shift of 343 and B cell of 312 (Table 2).

Figure 1.

Hematoxylin and Eosin staining of liver transplant biopsies. (A) POD 0—Small zones of confluent centrilobular necrosis are seen, similar to those changes commonly accompanying preservation/reperfusion injury. The dotted green line highlights the zone of confluent centrilobular necrosis, ×200, H&E, PA: portal area, CL: centrilobular zone. (B) POD 5—The biopsy shows centrilobular and midzone hepatocyte necrosis with focal panlobular necrosis with ductular proliferation, ×200, H&E. (C) POD 40—There is mild to moderate centrilobular hepatocanalicular cholestasis with ballooning of centrilobular hepatocytes. An accompanying fibroedematous expansion of portal areas with ductal proliferation is present, suggesting biliary tract outflow obstruction, ×200, H&E. (D) POD 64—Moderate centrilobular cholestasis with hepatocyte ballooning and fibrous portal expansion with ductular proliferation, a pattern suggesting progressive worsening of biliary tract outflow obstruction, ×100, H&E. (E) POD 104—There is centrilobular hepatocyte dropout and fibrosis. Two large bile plugs (upper center) are present. There is bridging fibrosis of portal areas with ductular proliferation, accompanied in this biopsy by a mild lymphocytic infiltrate. A mild superimposed acute cellular rejection cannot be histologically excluded, ×100, H&E. Each figure contains a high-magnification inset (×400) demonstrating the composition of the portal inflammatory cell infiltrate.

Figure 3.

Immunofluorescence for C4d staining. (A) POD 0 Biopsy with C4d positive immunofluorescence of sinusoids and centrilobular veins. (B) POD 40 Biopsy with C4d positive immunofluorescence of sinusoids. (C) Control biopsy from a patient with significant ischemic reperfusion injury with negative crossmatch and no detectable alloantibodies at time of liver reperfusion showing no significant C4d immunofluorescence. (D) Control biopsy from a patient with acute cellular rejection and without alloantibodies. Panel D shows nondiagnostic mild C4d immunofluorescence in periportal field.

Table 2.  Representation of the postoperative donor specific antibody testing and therapeutic interventions
Postoperative dayTreatmentCDC-AHG crossmatchFlow cytometry crossmatchHLA B7 Antigen beads
  1. 1The strength of donor-specific antibody reactivity is represented by the numbers in parentheses which are the ratio of the mean channel value of the B7 beads tested with patient serum divided by the mean channel value of the B7 beads tested with a negative control serum.

0 Positive T cell dilution 1:1Positive T cell channel shift 343T cell AHG-CDC 25% (B*7,51,65)
1 Negative T cellNegative T-cell channel shift 18 
8 Positive T-cell channel shift 464 
11IVIG (1 g/kg) 
12IVIG (1 g/kg) 
18 Positive T-cell channel shift 401 
28IVIG (1 g/kg) 
29IVIG (1 g/kg) 
34Rituximab (375 mg/M2) 
40 HLA-B7 flow bead positive (2.08)1
69 HLA-B7 flow bead positive (3.09)1
74 HLA-B7 flow bead positive (3.52)1
104IVIG (0.5 g/kg) HLA-B7 flow bead positive (3.06)1

The patient remained hemodynamically stable, but due to a persistent coagulopathy and hemoperitoneum required reexploration on POD 1. A repeat biopsy and closure of her fascia were performed on POD 5 after resolution of the coagulopathy. This POD 5 biopsy showed extensive centrilobular and midzone necrosis with focal panlobular necrosis (Figure 1B). There was a persistent elevation in the bilirubin and gamma-glutamyl transferase (GGT) (Figure 2). Doppler duplex ultrasonography of the hepatic vessels was normal and cholescintigraphy (HIDA) scan demonstrated no biliary leak or obstruction. Repeated flow crossmatches increased in positivity (Table 2). She was treated with 60 g of IVIG and responded with a decrease in bilirubin and alanine transaminase (ALT) (Figure 2). The patient was discharged to a rehabilitation unit 13 days after transplant on tacrolimus, mycophenolate mofetil and prednisone.

Figure 2.

Graph representing postoperative clinical course, diagnostic studies and therapeutic interventions. C4d positive biopsies (Bx): IVIG, plasmapheresis and rituximab are indicated.

On POD 15, the patient was readmitted with respiratory failure. She had a persistently elevated alkaline phosphatase and GGT (Figure 2). Doppler duplex ultrasonography and HIDA scans were again normal. The patient demonstrated persistent DSA and was treated again with 60 g of IVIG on POD 28 and 29 and again responded with decreased levels of serum bilirubin and ALT. On POD 34, rituximab 800 mg (375 mg/m2) was given intravenously. On POD 40, a biopsy was obtained due to increasing serum transaminases. It revealed hepatocanalicular cholestasis, necrosis of scattered lobular hepatocyte, fibroedematous expansion of the portal areas with neutrophils focally invading the wall and lumen of proliferating ductules. There was no evidence of acute cellular rejection (Figure 1C). C4d immunofluorescence was found with the same staining pattern as observed in the 1-h post-implantation biopsy with diffuse strong linear C4d deposits along the sinusoidal spaces and central veins (Figure 3B). Both extrahepatic and diffuse intrahepatic biliary ductal system abnormalities were excluded by repeated Endoscopic Retrograde Cholangiography (ERC) on POD 40 and 57.

Due to rapidly rising bilirubinemia, transaminasemia, and an increase in DSA (Figure 2 and Table 2), a liver biopsy was repeated on POD 64. This biopsy revealed cholestasis and bile duct proliferation with fibrous portal expansion (Figure 1D) similar to changes seen at POD 40. The patient received four plasmapheresis treatments with a steroid pulse in an attempt to rescue the liver graft by controlling the rising DSA. She had a temporary improvement in her serum transaminases and bilirubin, but her DSA rebounded rapidly. On POD 104, due to increasing serum transaminases and bilirubin, a repeat biopsy was obtained which showed cholestasis, ductular proliferation and bridging and centrilobular fibrosis. In addition, mild lymphocytic infiltrates were found for the first time in portal areas suggesting that mild acute cellular rejection could not be entirely excluded (Figure 1E). Serology again demonstrated DSA and the patient was treated with 42 g of IVIG.

As a consequence of the above course and poor graft function, the patient was offered repeat liver transplantation, but she refused. Her pulmonary insufficiency worsened. The patient did not wish to be reintubated in this clinical setting and died on POD 114 of respiratory failure. The family declined a postmortem examination.

Response to thymoglobulin and rituximab

Thymoglobulin induction resulted in an absolute lymphocyte count less than or equal to 0.1 × 109/L for longer than 8 days posttransplant. Rituximab given on POD 34 decreased CD19 and CD20 counts to undetectable levels when measured 36, 54 and 72 days after infusion.

Tissue typing and alloantibody determination/characterization

The recipient's HLA Antigens were: A*11,30; B*13,27; DRB1*04,07. The donor's HLA Antigens were: A2,31; B7,45; DR4,15.

The recipient's panel reactive antibody (PRA) was 25% by standard T-cell AHG-CDC cell tray analysis on pretransplant serum and demonstrated reactivity to the HLA-B7, 51 and 65 antigens. Flow PRA of this pretransplant serum was 97% for Class I reactivity.

Postoperatively, the patient had persistent DSA to the Class I B7 antigen (Figure 4) despite attempts to remove antibodies by plasmapheresis, intravenous immunoglobulin and rituximab (Table 2).

Figure 4.

Flow PRA Class I single antigen beads analysis of serum on POD 40. Recipient serum had strong reactivity against donor B7 antigen as well as additional third party antigens. The no antigen beads served as negative control. The B27 recipient's antigen beads did not show reactivity.


The diagnosis of AMR after kidney or heart transplantation requires four features as proposed by a national conference held at the NIH to assess current knowledge regarding humoral solid organ rejection. In their 2003 meeting report, these features of AMR include: (1) clinical evidence of graft dysfunction, (2) morphologic evidence of tissue injury, (3) immunopathologic evidence for antibody mediated response, i.e. anatomically specific C4d deposits and (4) serologic evidence of circulating antibodies to donor HLA or other donor antigens (8).

In liver transplantation, there have been previous reports of presumed humoral rejection when using ABO compatible donors (6,7). We report the first case of AMR in a liver transplant recipient that includes a complete scenario of: (1) a clinical course of chronic cholestasis without common bile duct obstruction, with (2) histological tissue injury starting with periportal edema and centrilobular necrosis and progressing to fibrosis without significant acute cellular rejection, (3) specific linear C4d immunofluorescence throughout the sinusoidal endothelium performed on frozen tissue obtained both shortly after reperfusion as well as 40 days later, (4) persistent circulating alloantibodies to donor specific HLA and additionally, we observe (5) temporally associated clinical responses to similar therapies used for treatment of renal humoral rejection.

AMR is accepted as a separate clinical and pathological entity in kidney and heart transplantation. The inability to directly demonstrate antibody deposits in renal biopsies has been partially overcome by the availability of an immunofluorescent stain for the complement degradation product C4d. Investigators correlated the site-specific peritubular capillary deposits of C4d with detection of anti-HLA DSA. The specificity of C4d as a marker has been validated in studies (11).

C4d staining and the documentation of circulating DSA in the presence of renal graft dysfunction is now accepted by most clinicians as evidence of active humoral rejection in renal and heart transplantation. In kidneys, immunofluorescence presents a specific pattern of peritubular capillary C4d deposition. Polymorphonuclear or mononuclear cells may also be observed in peritubular capillaries. Clinical attempts to remove circulating antibodies often results in improvement of renal function with resolution of C4d deposits in repeat biopsies. Renal antibody-mediated injury has been controlled using combinations of plasmapheresis, mycophenolate mofetil, tacrolimus, IVIG, polyclonal antilymphocyte globulin and anti-CD20 monoclonal antibody (rituximab) (12).

C4d staining in kidney, heart, lung and pancreas allografts undergoing AMR is consistently demonstrated along the microvasculature of the involved organ. In heart transplantation, a linear C4d staining pattern is observed along the myocardial capillaries (8). A recent report demonstrates linear C4d staining in the interacinar capillaries during a clinical case of AMR in a pancreas allograft (13). Lung allografts undergoing AMR have also shown C4d linear positivity in capillaries and arterioles (14). Consistent with Colvin's editorial, we propose that AMR in a liver graft demonstrates linear C4d staining throughout the dominant capillary bed, i.e. the sinusoidal endothelium (5).

In both ABO compatible and incompatible liver transplantation, there are reports of C4d staining along the endothelium of portal venule and arteries. Previous reports of C4d immunostaining do not include serological data or frozen section immunofluorescence microscopy employing a monoclonal antibody staining directed against C4d (15–19). Based on our own experience in renal allografts, C4d staining on frozen tissue samples with monoclonal antibodies gives the most specific and reproducible results. We believe that the staining profile we observed in the patient under discussion is specific. There is a lack of consensus as to the anatomic pattern or interpretation of C4d staining in acute rejection in liver transplantation (5). No reports describe a strong linear C4d staining pattern throughout the entirety of the sinusoidal endothelium. A report involving an ABO compatible liver transplant recipient undergoing AMR by Moeller et al. (20) describes some capillaries of the liver sinus showing “a partially focal and partially diffuse accumulation of C4d”. Sawada et al. (17) describe C4d staining in some liver sinusoids during moderate and severe cellular rejections.

As others have postulated (5,18), the periportal C4d staining observed by some groups may be “nonspecific”. Periportal C4d staining parallels nonspecific C4d staining in the kidney in arterioles with sclerosis or along atrophic tubular basement membranes. Nonspecific activation of complement resulting in C4d staining may be due to the activation of innate immunity triggered by cellular infiltrate due to rejection, bacterial infection or viral infection, including recurrent viral hepatitis. In our case, we have multiple biopsies negating acute periportal cellular infiltration throughout the majority of the transplant course. We believe that our case presents isolated AMR due to preformed and ongoing production of circulating DSA. Initial lack of lymphocytic infiltration leads us to speculate that the release of soluble HLA antigens from the liver graft was responsible for an ongoing HLA alloantibody production in the host and worsening of graft function.

The clinical picture and histology share some similarities to humoral injury in ABO incompatible liver transplantation reported by others (2,3). We did not identify all the typical morphological features reported in acute humoral rejection of liver allografts from ABO incompatible donors in this patient. We did not observe sinusoidal infiltrates of neutrophils, fibrin, and focal hemorrhage. Typical humoral rejection findings that were observed in our case report include initial centrilobular necrosis, similar to those changes commonly accompanying preservation/reperfusion injury (2) followed at POD 40 with portal and periportal edema and hepatocyte necrosis (3). These later histologic features that resemble biliary tract outflow obstruction prompted us to repeatedly attempt to refute this diagnosis through multiple ERC's and HIDA scans. The extensive and probably localized panlobular coagulative necrosis identified in the POD 5 biopsy (Figure 1B) is likely a manifestation of severe ischemic reperfusion injury parallel with acute AMR. With increasing time posttransplantation, progressive portal fibrosis developed. It is probable that detection of AMR evolving in patients with liver allografts from ABO compatible donors is challenging to recognize on an H&E section alone with features in common with ischemic reperfusion injury. The evolution of persistent histologic features of biliary tract outflow obstruction in the absence of clinical and radiographic evidence for large duct obstruction or ischemic intrahepatic biliary strictures may be an indicator of AMR.

In conclusion, we propose that the humoral alloresponse in ABO compatible liver allografts may cause significant graft injury without acute cellular rejection. AMR is demonstrated by persistent DSA and a distinctive strong linear C4d staining along sinusoidal and central vein endothelium by immunofluorescence microscopy. We believe histologically AMR initially mimics severe ischemic reperfusion injury and may later be associated with periportal edema, necrosis and late bridging fibrosis in a pattern that can mimic biliary outflow obstruction. A systematic analysis of a large population of post-liver transplantation C4d immunofluorescence biopsies with corresponding serology data and clinical outcomes is needed. This may help to elucidate the role and significance of humoral alloresponse after liver transplantation.