1. Top of page
  2. Abstract

The importance of antibody-mediated rejection (AMR) in ABO-compatible liver transplantation is controversial. Here we report a prospective series of liver recipients with a preoperative positive crossmatch. To establish the diagnosis of AMR in liver recipients, the criteria described for kidney allografts were adopted. In approximately 10% of 197 liver transplants, we observed a positive T and B cell flow crossmatch before transplantation. Fifteen of 19 patients converted to negative crossmatches early after transplantation and displayed normal liver function while they were on routine immunosuppression. Four patients maintained positive crossmatches. Three of the 4 met the criteria for AMR and showed evidence of graft dysfunction, the presence of donor-specific antibodies (DSAs), morphological tissue destruction with positive C4d linear staining on the graft sinusoidal endothelium, and improved function with attempts to eliminate DSAs. A persistently positive crossmatch after liver transplantation may lead to early, severe AMR and liver failure. C4d staining in the liver sinusoidal endothelium should alert one to the possibility of AMR. In our experience, patients with a positive crossmatch should have it repeated at 2 weeks and, if it is positive, again at 3 to 5 weeks. Recipients with an unknown preoperative crossmatch who develop early cholestasis of unclear etiology should be crossmatched or tested for the presence of DSAs to evaluate for AMR. Liver Transpl, 2011. © 2011 AASLD.

The significance of antibody-mediated rejection (AMR) in ABO-compatible liver transplantation is still controversial. There are conflicting reports concerning short- and long-term outcomes in the setting of a positive human leukocyte antigen (HLA) crossmatch.1-8 To date, only 1 case report has documented the presence of donor-specific antibodies (DSAs), graft dysfunction with histological findings of liver AMR, and the presence of C4d staining in liver sinusoids.1 The combination of these characteristics is absent from other reports on liver AMR.9-10

Case reports of hyperacute rejection11 and series of positive crossmatch (PC) liver transplants11-13 have made the transplant community aware of the poor outcomes possible for patients undergoing transplantation across a PC. Hyperacute rejection is a rare occurrence in liver transplantation,2 even with a PC, and is not included in acute AMR, as discussed in this article.

Early AMR injury of the liver tissue is often unrecognized as an immunological event. The lack of clear clinical, histological, and immunohistological criteria for diagnosing AMR and the absence of crossmatch results in many liver recipients may contribute to the underreporting of AMR in liver recipients versus recipients of other solid organ transplants. The prevalence of a PC or a high panel reactive antibody (PRA) level (a surrogate for a PC) in recipients ranges from 7% to 33%.12, 14 There are reports indicating the poor performance of liver grafts transplanted into patients with a PC13 and/or a high PRA level with reduced 1-year graft and patient survival, but there has been no demonstrable effect on long-term results.14

An additional factor that may add to the confusion in recognizing AMR is that the earliest clinical manifestation of this immunological event is cholestasis, a nonspecific clinical and pathological finding.15 This clinical presentation may curtail alertness for AMR because cholestasis is observed in a wide variety of common conditions after transplantation, such as ischemia/reperfusion injury, initial graft dysfunction in cases of donor liver steatosis, biliary obstruction, and/or stenosis, hepatic artery thrombosis, acute cellular rejection (ACR),3 sepsis, ABO incompatibility, viral infections, recurrent primary disease,15 nonanastomotic biliary strictures related to initial posttransplant bile cytotoxicity,16 and the presence of a preformed non-HLA antibody leading to biliary injury and rejection.17 Similarly, the histological findings for AMR (proliferation of small bile ducts, centrilobular hepatocyte swelling and single cell necrosis, sinusoidal accumulation of neutrophils, and hepatocanalicular cholestasis) are not specific and can also be found in preservation reperfusion injury.12, 18 This necessitates additional evidence, such as the presence of DSAs and linear C4d staining, to make an unequivocal diagnosis of AMR.

In 2006, our group published a well-documented case of isolated AMR without a component of cellular rejection.1 The reported patient's graft biopsy samples demonstrated clear linear immunofluorescence (IF) staining for C4d in liver sinusoids of a frozen section specimen. In this case, morphological tissue injury, the presence of DSAs in serum, and graft dysfunction were observed. In addition, intermittent improvement of liver graft function was noticed with reductions in DSA levels after the initiation of various antibody-reducing treatments [plasmapheresis (PP), intravenous immunoglobulin (IVIG) infusion, and rituximab]. These criteria, including linear sinusoidal C4d deposits, are identical to those used to define AMR in pancreas, lung, heart, and kidney transplantation.19-24

Here we report a case series with pretransplant positive crossmatches (PPCs) and describe the spectrum of clinical outcomes. We have attempted to identify which PC patients are at risk for graft dysfunction and loss. This report correlates the presence of circulating DSAs in a recipient's blood with the clinical presentation of AMR in the liver allograft and pathological features of antibody-mediated or humoral rejection, including positive linear staining for C4d on the sinusoidal endothelium.


  1. Top of page
  2. Abstract


From January 2005 to November 2008 at the University of North Carolina at Chapel Hill, we transplanted 197 livers into 182 recipients. Fifteen patients (10.7%) required retransplantation. Only 2 patients, both with PPC but without AMR, received a second liver transplant. The first patient underwent retransplantation on postoperative day (POD) 16, and the second underwent retransplantation a year after the initial liver transplant. Ten recipients were pediatric patients. We identified 19 patients (10%) with a PPC and a group of 19 patients with significant reperfusion liver allograft injury and a negative crossmatch (NC; Table 1). In this report, we focus mainly on the patients with a PPC against their liver donors. The crossmatch results were reported within 24 hours of the transplant event. All patients underwent transplantation with livers from deceased ABO-compatible donors. After PC determination, patients were followed prospectively with repeat crossmatches performed in week 2 and in weeks 3 to 5; they were then repeated as needed for persistent PC and DSA levels. Those diagnosed with AMR (D-AMR) were evaluated for biliary complications and ischemia due to hepatic artery stenosis or occlusion with endoscopic retrograde cholangiopancreatography, Doppler ultrasound, magnetic resonance scanning, and/or angiography. We routinely tested all patients for liver function tests every day in the hospital and at least weekly for the first 3 months, performed biopsy according to clinical indications, periodically tested every recipient for viral infections, and followed recipients for the potential recurrence of primary disease to exclude cholestasis due to factors other than rejection. None of the D-AMR patients displayed hepatic artery or bile duct abnormalities. The diagnosis of AMR was based on the presence of DSAs in the recipient serum, liver dysfunction with cholestasis, and histological features of AMR with positive staining for C4d in liver sinusoids. Liver dysfunction was defined as elevated levels of total bilirubin (TB),14 gamma-glutamyl transpeptidase (GGT), alkaline phosphatase, and/or alanine aminotransferase (ALT) and aspartate aminotransferase (AST). The routine immunosuppressive regimen consisted of treatment with tacrolimus, mycophenolic acid, and steroids. Steroids were tapered off for most patients by the end of the first year. The tacrolimus dose was tapered within the first 6 months to maintenance levels of 5 to 8 ng/mL. The initial trough levels were set at 8 to 10 ng/mL. As part of the routine, all patients with well-functioning grafts were on monotherapy 12 to 18 months after transplantation. Patients who underwent transplantation for liver failure related to autoimmune diseases were maintained on a triple therapy with tacrolimus, mycophenolic acid, and prednisone for life. The immunosuppressive therapy was augmented by an antibody-reducing treatment (PP, IVIG, and rituximab) in D-AMR patients after an attempted treatment with increased doses of the standard triple immunosuppression. All data were recorded in our computerized database and re-evaluated for the purpose of this publication.

Table 1. Characteristics of the D-AMR, ND-AMR, and NC Groups
CharacteristicPPC (n = 19)PNC (n = 19)P Value: PPC Versus PNC
D-AMR (n = 3)ND-AMR (n = 16)P Value: D-AMR Versus ND-AMR
  • NOTE: Two-sided P values (Fisher's exact test) are shown; P values < 0.05 are considered significant.

  • *

    One patient died on POD 7 because of primary nonfunction.

  • †, *

    One patient developed late biliary complications that were likely due to a postbiopsy arterial-venous fistula.

Age, years (mean ± SE)51 ± 2 52 ± 20.76
MELD score (mean ± SE)15.00 ± 4.9322.00 ± 20.1920 ± 10.86
21.2 ± 2
AST on day 1, U/dL (mean ± SE)2363 ± 6531685 ± 3700.47  
1792 ± 327 4827 ± 8800.003
Cold ischemia time, hours (mean ± SE)6.8 ± 0.4 6.7 ± 0.30.92
Time to arterial anastomosis, minutes (mean ± SE)101 ± 4 108 ± 80.39
 PPC (n = 19) PNC (n = 19)P Value: D-AMR Versus PNC
AMR CriteriaD-AMR (n = 3)ND-AMR (n = 16)P Value: D-AMR Versus ND-AMR
Strong linear C4d staining [n/n (%)]3/3 (100)1/16 (6.25)0.0040/190.0006
Persistent PC [n/n (%)]3/3 (100)1/16 (6.25)0.0040/190.0006
Improving function [n/n (%)]0/3 (0)15/16 (93)0.0118/19 (95)*0.0026
All criteria for AMR [n/n (%)]3/3 (100)0/16 (0)0.0010/19 (0)0.0006

HLA Typing, Antibody Screening, and Crossmatching

Patient and confirmatory donor HLA typing (HLA-A, HLA-B, and HLA-DRB1) was determined by polymerase chain reaction with a sequence-specific primer (Invitrogen, Carlsbad, CA) or polymerase chain reaction with sequence-specific oligonucleotide probes (LabType, One Lambda, Inc., Canoga Park, CA). Alloantibody profiles were assessed by solid-phase testing with flow cytometry or a multiplex bead array. All sera were screened for class I and II HLA antibodies with FlowPRA class I and II screening beads (One Lambda). Positive screening tests were further tested with FlowPRA single antigen beads or LabScreen single antigen beads (One Lambda) to define antibody specificities. All data were normalized to negative control serum and were reported as normalized baseline values [called the mean fluorescence intensity (MFI)]. A normalized baseline value greater than 1000 was considered positive.

Flow cytometry crossmatches against pronase-treated donor lymphocytes were performed with a 3-color technique [anti–IgG F(ab′)2 fluorescein isothiocyanate, CD19PE, and CD3PerCP] with a FACSCalibur flow cytometer (BD Biosciences, Inc., Rockville, MD). Positive results were defined as a channel shift greater than 40 for T lymphocytes and as a channel shift greater than 50 for B lymphocytes and are reported as medians and standard deviations.

Patients with positive T and B cell crossmatches determined by flow cytometry were monitored with repeat crossmatches until donor cells were exhausted. DSAs were determined on pretransplant sera and monitored after liver transplantation to establish an AMR diagnosis and to assess the progress of the rejection treatment. After the determination of anti-HLA alloantibodies, the calculated panel reactive antibody (CPRA) level was determined with the United Network for Organ Sharing formula to evaluate the extent of a recipient's sensitization.


Liver core biopsy tissues were fixed in 10% buffered formalin, and 5-μm sections were cut from the paraffin-embedded tissue for hematoxylin and eosin (H&E) and trichrome stains for morphological evaluation. H&E slides were evaluated on the basis of previously described pathological criteria for liver AMR.12 C4d IF staining was performed on 4-μm-thick sections cut from tissue rapidly frozen in precooled isopentane. A mouse monoclonal antibody (Quidel, San Diego, CA) was used for the detection of C4d as described in a previously published protocol.25 The results of the C4d IF staining involving sinusoids were recorded semiquantitatively. Diffuse staining was defined as involving >50% of the sinusoidal compartment, whereas focal staining involved <50% of the sinusoidal compartment. The pattern of C4d deposition within sinusoids was described as either linear or granular. The intensity of the staining was recorded as follows: (0) negative, (1+) weak, (2+) moderate, and (3+) strong. Periportal C4d deposition in the endothelium of portal vessels (veins and capillaries) or arteries was also documented. Specimens were evaluated independently by 2 pathologists. The pathologists were not aware of the clinical diagnosis or treatment.

Statistical Analysis

Results are expressed as means and standard deviations unless otherwise stated. Fisher's exact test was used to compare proportions. The 2-tailed Student t test was performed for the comparison of groups with parametric data. One-way analysis of variance with Tukey's multiple comparison test was used to compare TB levels between the 3 groups over time (GraphPad Software, La Jolla, CA). A 2-sided P value < 0.05 was considered statistically significant.


  1. Top of page
  2. Abstract

We identified 19 patients (approximately 10%) with a PPC (Table 1). Two of them had cirrhosis secondary to alcohol, 4 were diagnosed with cryptogenic cirrhosis, 6 were diagnosed with hepatitis C, 1 was diagnosed with hepatitis B, 3 were diagnosed with primary biliary cirrhosis, 2 were diagnosed with primary sclerosing cholangitis, and 1 was diagnosed with autoimmune hepatitis. The observation time after transplantation ranged from 1.3 to 47 months (mean = 19.8 ± 12.7 months).

Applying the clinical and histological criteria listed in the introduction, we divided the PC recipients into 2 groups. Patients in the first group (the D-AMR group; n = 3) met all the criteria previously described (Table 1). They all received augmented immunosuppressive treatments, such as PP, IVIG, rituximab, thymoglobulin (Thymo), and/or OKT3. Patients in the second group did not meet all the AMR criteria [ie, the group not diagnosed with AMR (ND-AMR); n = 16] despite the presence of a PC on the serum sample collected before liver transplantation. In the ND-AMR group, 15 of 16 patients reversed their crossmatches on routine immunosuppression within 2 weeks. Only 1 patient (ND-1) received 1 dose of rituximab on POD 32 for persistent PC. This patient was crossmatch-positive 12 months after transplantation but became crossmatch-negative and cleared all 4 DSAs when she was tested 24 months after liver transplantation. Three patients died: 2 in the D-AMR group 116 and 40 days after liver transplantation (D-1 and D-2, respectively) and 1 in the ND-AMR group 62 days after liver transplantation (ND-8). All 3 deaths in the groups were due to pneumonia and/or sepsis with poorly functioning liver grafts. One patient (D-3) in the D-AMR group successfully underwent retransplantation.

The third group of patients had NCs and severe reperfusion injury to their liver allografts (as demonstrated by AST values measured on POD 1). One patient died on POD 7 from primary graft nonfunction. This group displayed the highest level of transaminasemia (Table 1) but no positive sinusoidal C4d staining (Table 2) and no evidence of DSAs.

Table 2. Histology of D-AMR Patients and Selected ND-AMR and NC Patients
PatientBiopsy DayAntibody-Mediated Injury H&E*C4d IF*RAI
Cholangiolar ProliferationCentrilobular Hepatocyte SwellingSinusoidal NeutrophilsHepatocanalicular CholestasisSinusoidalPeriportal
  • *

    The following scale was used: (0) no change, (1) mild, (2) moderate, and (3) strong.

  • The Banff schema (0-9) was used for grading liver allograft rejection.

 D-1POD 00necrosis303 (linear)00
POD 52necrosis21  2
POD 4021113 (linear)02
POD 643202  2
POD 1043103  2
 D-2POD 0, back table0necrosis20000
POD 0, reperfused0necrosis313 (linear)00
POD 723313 (linear)00
POD 1122022-3 (linear)00
POD 2732131 (focal)00
 D-3POD 01020  0
POD 730313 (linear)2 (granular)6
POD 1432323 (linear)2 (granular)5
POD 2421213 (linear)03
POD 422221  2
POD 8622111 (focal) 6
POD 1092011  4
 ND-1POD 3311103 (linear)02
 ND-2POD 00131  0
 ND-3POD 01311  0
 ND-4POD 00120  0
 ND-5POD 211100  0
POD 46021111 (linear)00
 ND-6POD 00221  0
POD 181132  0
 ND-7POD 00100  0
 ND-8POD 00111  0
POD 1022222 (linear)02
 ND-12POD 11221  0
 NC-1POD 00000000
 NC-2POD 0000001 (granular)0
 NC-3POD 00000000
 NC-4POD 00020000
 NC-5POD 0001001 (granular)0
 NC-6POD 00010000
 NC-7POD 00010000
 NC-8POD 0002001 (granular)0
 NC-9POD 0003001 (granular)0

Clinical Course (Figs. 1 and 2)

The 3 recipients in the D-AMR group all experienced significant damage to their liver grafts while they were on maintenance immunosuppression; this was manifested by high levels of serum TB, alkaline phosphatase, GGT, ALT, and AST. Patients in this group did not respond to increased levels of standard immunosuppression or intravenous steroid boluses, but they did respond to treatment of their AMR with IVIG, rituximab, PP, Thymo, and/or OKT3 with transient improvements in liver function tests. Patient D-2 (Fig. 1B) underwent treatment with Thymo (1.5 mg/kg) on PODs 1 to 4, 5 sessions of PP, 3 infusions of IVIG (1 g/kg/day), and rituximab. This patient's DSAs did not respond to extensive therapy, and the liver function parameters progressively worsened. Immunosuppression was reduced on POD 31 when the patient developed enterococcal sepsis, which led to death on day 40. Patients D-1 (Fig. 1A) and D-3 (Fig. 1C) also required antibody-depleting therapy, and they responded with transient reductions of DSAs followed by transient improvements in liver function (Figs. 1 and 2). Both patients were offered retransplantation because of chronic severe graft dysfunction. Patient D-1 refused a second liver transplant and expired because of exacerbation of hepatopulmonary syndrome caused by poor liver graft function, pneumonia, and then sepsis. D-3 successfully underwent retransplantation 13 months after his initial graft failed because of AMR. The failed allograft displayed typical end-stage features as well as weakening C4d sinusoidal staining and chronic rejection after repeated treatments for AMR and ACR. His immunosuppressive medications had been tapered before retransplantation to improve his chances of avoiding severe infection during retransplantation.

thumbnail image

Figure 1. Clinical course and TB, ALT, and GGT responses to treatment. For clarity, the data for 3 patients of the D-AMR group (D-1, D-2, and D-3) and for 1 patient of the ND-AMR group (ND-1) are presented. ND-1, despite having a PC within the first 2 postoperative weeks and positive C4d IF in liver sinusoids, had good liver function without cholestasis. The scales have been adjusted for each patient to better reflect fluctuations of laboratory values. Treatment with IVIG, PP, rituximab, Thymo, and biopsy days are indicated. The diagnosis of AMR was made on PODs 7 to 10 after the observation of features of cholestasis, sustained PC, an H&E pathological evaluation, and strong linear deposits of C4d in liver sinusoids detected by IF in frozen specimens. A transient improvement of cholestasis was observed after the treatment for AMR. The first 42 days of the course are shown.

Download figure to PowerPoint

thumbnail image

Figure 2. Comparison of the TB levels of the 3 patient groups. In the ND-AMR and NC groups, the TB levels normalized after 3 weeks. Patients in the D-AMR group showed transient improvements in their TB levels at 3 weeks, which were likely related to PP, IVIG, and/or cell-depleting therapy (T cell, B cell, or both; Fig. 1), with subsequent rebound. The TB curve for the D-AMR group is different from the curves for the ND-AMR and NC groups (P = 0.013). The NC group showed no difference versus the ND-AMR group (P > 0.05). The results are presented as means and SEs (SEs were used for better presentation on the graph).

Download figure to PowerPoint

D-AMR patients were screened with Doppler surveillance testing of the liver on multiple occasions; this testing started on POD 0 and continued throughout their course of observation. For standard clinical indications of graft dysfunction, we used hepatic angiography, endoscopic retrograde cholangiopancreatography, and/or percutaneous cholangiography to rule out hepatic artery thrombosis or biliary stenosis. In this series, no patient in the D-AMR group had any vascular or biliary abnormalities detected.

Fourteen of 16 patients in the ND-AMR group achieved stable liver allograft function within the first 4 weeks after transplantation. Patients ND-1 (Fig. 1D), ND-12, ND-5, and ND-2 had good liver function within the first week after transplantation; ND-6, ND-13, ND-14, and ND-7 had good liver function by week 2; and ND-3, ND-8, and ND-4 had good liver function within the first 4 weeks after transplantation. Patient ND-9 maintained elevated TB, GGT, AST, and ALT levels for the first 13 weeks; then, the graft function parameters normalized. The liver function of patient ND-6 deteriorated 9 weeks after transplantation. A series of investigations revealed an anastomotic biliary stricture and an arterial-venous fistula (likely from the previous biopsy). Liver function improved after endoscopic interventions in posttransplant week 35, but mild cholestasis persisted. For all patients in the NC group (except for 1 patient who died on POD 7), their liver function stabilized within the first 2 to 3 weeks. Posttransplant normalization of liver function tests in the ND-AMR and NC groups occurred according to similar patterns and at similar times, as indicated in Fig. 2, which presents a comparison of TB levels. The TB levels and dynamics of normalization in the ND-AMR and NC groups were significantly different from those in the D-AMR group.

Analysis of CPRA, Crossmatching, and DSAs


The CPRA level was 99.6% ± 0.6% for patients in the D-AMR group and 82% ± 23.8% for patients in the ND-AMR group (P = 0.27).


T and B cell flow cytometry crossmatching was performed on samples obtained preoperatively on the day of transplantation, within the first 2 weeks after liver transplantation, and within 3 to 5 weeks after transplantation with stored lymphocytes from the liver donors. Only patients in the D-AMR group were followed with T and B cell crossmatching beyond posttransplant week 6. In most cases, the lack of donor cells limited our ability to continue crossmatch follow-up. Subsequently, solid-phase DSA detection was used. Pretransplant and repeated follow-up crossmatches in the D-AMR group remained strong throughout the entire time of observation (Fig. 3), with T cell crossmatch results of 366 ± 219, 198 ± 153, and 251 ± 46 channel shifts in the 3 described time periods. B cell crossmatch results were similar: 349 ± 132, 264 ± 115, and 391 ± 2 channel shifts, respectively. Among all 19 patients with a PPC, only 1 patient (D-1) had a positive T cell crossmatch and a negative B cell crossmatch. All other patients had positive T and B cell crossmatch tests on serum collected before surgery on the day of transplantation.

thumbnail image

Figure 3. T cell flow cytometry crossmatch results. B cell crossmatch results are not presented but showed a pattern similar to that of T cells. In the D-AMR group, all patients maintained a strong PC after surgery. Patients of the ND-AMR group reversed their T and B cell crossmatches spontaneously within the first 2 weeks after liver transplantation. The ND-AMR patients converted their crossmatch results while they were on routine immunosuppression (tacrolimus, mycophenolic acid, and prednisone). Patient ND-1 (ND-AMR) converted her crossmatch to a low positive on POD 2, but on POD 7, her T and B tests increased to 436 and 394 channel shifts, respectively. Donor cells for this patient lasted until POD 18. The T and B cell channel shifts were 401 and 343, respectively. Thereafter, the patient was followed by repeated DSA testing. ND-1 had normal liver function, did not meet our criteria for AMR, and was assigned to the ND-AMR group.

Download figure to PowerPoint

In the ND-AMR group, the median channel shift of the T cell crossmatch before surgery was 230.6 ± 38. During the first 2 weeks after liver transplantation, the mean channel shift was reduced to 36.9 ± 14.6 (negative) and only partially recovered between weeks 3 and 5 after surgery for a mean channel shift of 91.4 ± 96.7 (Fig. 3). The B cell crossmatch was 229 ± 115, 28.1 ± 12.6, and 99.4 ± 89.5 channel shifts before transplantation and in the respective weeks of follow-up. The reduction of the intensity of the T and B crossmatches was spontaneous in the ND-AMR group and was not a result of a specific treatment leading to alloantibody reduction (except for patient ND-1, who received 1 dose of rituximab).

DSAs and Third-Party Alloantibodies

Each recipient of D-AMR had 3 to 7 identifiable DSAs. The first patient (D-1) had 3 DSAs identified, the second patient (D-3) had 4 DSAs identified, and the third patient (D-2) had 7 DSAs identified (Fig. 4A). All of them maintained a high or increasing level, with channel shifts of single antigen beads ranging from 4982 to 14,174 before transplantation and from 4779 to 16,898 by the end of the observation period. DSA levels fluctuated in response to PP, rituximab, and IVIG but remained elevated over the long term (Fig. 4A). In 11 patients from the ND-AMR group, the DSA levels were undetectable or were detectable at significantly reduced levels after transplantation in comparison with pretransplantation values. Three patients reduced their DSAs to only 1 specificity, and 1 patient reduced his DSAs from 4 to 2 within the first 13 weeks of observation. Subsequent follow-up revealed a reduction of DSAs to very low levels (below 1000 channel shifts for single antigen beads; Fig. 4B). Two patients (ND-1 and ND-6) maintained 1 DSA specificity with a channel shift strength of 4200 to 13,368 for more then a year. One of them (ND-1) cleared DSAs completely between 12 and 24 months. Fourteen of 16 ND-AMR patients cleared their DSAs within 4 weeks of liver transplantation (Fig. 5). In 50% of the liver allograft recipients, third-party alloantibodies also became undetectable within the first 12 months, and in some patients, they reoccurred periodically.

thumbnail image

Figure 4. (A) DSA dynamics in patient D-2 (D-AMR). Fluctuations in the DSA channel shifts within a 4-week period reflect the patient's treatment with rituximab, IVIG, and PP. This patient nearly eliminated 2 class II DSAs, but 4 other DSAs remained strong throughout the course of follow-up. (B) Six-week follow-up of the DSA channel shifts in patients of the ND-AMR group: ND-1 (A2, A66, B57, and B58), ND-2 (A11 and DR7), ND-3 (B7), ND-4 (B18, B65, C7, and C8), ND-5 (A3, A26, B45, and DR4), ND-6 (A2, A24, and B7), ND-7 (DR13 and DR17), ND-9 (A2, A11, DR4, and DR53), ND-10 (A68, B53, and B58), ND-11 (A11), and ND-12 (B7 and B35).

Download figure to PowerPoint

thumbnail image

Figure 5. Number of sustained DSAs after transplantation. In the D-AMR group, liver recipients maintained their positive flow crossmatches. Crossmatches corresponded with the presence of DSAs during follow-up. Despite efforts to eliminate the antibodies, only 1 or 2 DSAs were either eliminated or reduced significantly in patients' sera. In the ND-AMR group, the conversion of the crossmatch coincided with the elimination of DSAs. Patient ND-1 cleared 3 DSAs within the first 12 months. Her last DSA (A-2 specificity) was still at the 3000 channel shift on single antigen beads in month 12. It was no longer detectable in month 24.

Download figure to PowerPoint

Pathology and C4d Staining


Patients in the D-AMR group had moderate to strong progression of histological changes related to antibody-mediated injury: proliferation of small bile ducts, centrilobular hepatocyte swelling and single cell necrosis, sinusoidal accumulation of neutrophils, and hepatocanalicular cholestasis. Only 1 patient (D-3) was diagnosed with concurrent ACR on POD 14 [rejection activity index (RAI) = 5/9]. The other 2 patients had isolated AMR. The ND-AMR group displayed only mild to moderate proliferation of small bile ducts, centrilobular hepatocyte swelling and single cell necrosis, sinusoidal accumulation of neutrophils, and hepatocanalicular cholestasis. Recipients in the NC group showed minimal changes on H&E that were isolated to only sinusoidal neutrophil deposits of a mild to moderate scale. Biopsy samples from this group (presented in Table 2) were collected 1 to 2 hours after liver reperfusion on POD 0. None of these patients had clinical indications for follow-up liver biopsy within the observation period. All initial C4d-positive biopsy samples in the D-AMR group were collected before treatment with any therapeutic antibody. We did not use an induction antibody in our liver transplant protocol.

C4d IF

Two specimens were collected from the normal liver parenchyma of 2 nontransplant individuals who underwent resection of benign liver lesions (hemangioma and adenoma). Nineteen specimens were collected from back-table biopsies before transplantation. None of these specimens revealed sinusoidal linear staining by IF. In 7 specimens (33%), which included the 2 nontransplant samples, IF staining revealed moderate to strong staining of arteriolar internal elastic lamina. Periportal stroma staining was noted in 8 biopsy samples (38%) among these control specimens. In 1 nontransplant individual, IF staining was evident in terminal hepatic venules.

An IF evaluation of biopsy samples collected from transplant patients with PCs and NCs (multiple biopsy samples for PC patients) showed variable staining of internal elastic lamina and portal stroma regardless of the pretransplant crossmatch. Only sinusoidal staining was characteristic for livers perfused with blood containing DSAs. The difference in sinusoidal staining between the PC and NC groups was statistically significant (Table 3).

Table 3. C4d IF Staining of Biopsy Samples From Patients With PPCs and NCs
 NC (n = 19)PPC (n = 18)P Value
  1. NOTE: Two-sided P values (Fisher's exact test) are shown.

Sinusoids1 (focal/granular)1816 (all linear, >2+)2<0.0001
Internal elastic lamina1636110.45
Portal stroma1094140.10

In all 3 patients in the D-AMR group, IF staining for C4d revealed strong linear deposits along the sinusoidal endothelium (Fig. 6 and Tables 1 and 2). The C4d positivity of patient D-3 was identified in a specimen with concurrent ACR (RAI = 6/9). The ACR component improved with Thymo treatment within 3 weeks; persistent AMR was found in subsequent biopsy samples with positive C4d staining. For patient D-2, negative IF staining of a preimplantation back-table biopsy sample was converted into strong linear sinusoidal staining after liver reperfusion on POD 0. Recipient ND-1 had strong linear staining for C4d but only mild histological findings on H&E (Table 2).

thumbnail image

Figure 6. IF for C4d staining: biopsy sample from patient D-2. (A) Positive IF of sinusoids and centrilobular veins on POD 7. (B) Negative IF in the biopsy sample taken from the donor liver during back-table procurement.

Download figure to PowerPoint

Seven patients from the NC control group were biopsied and examined for C4d after the day of transplantation. Patient NC-10 was negative for C4d on PODs 0, 3, and 136. NC-11 was negative on day 780, NC-12 was negative on PODs 0 and 52, NC-13 was negative on POD 3, NC-14 was negative on PODs 0 and 134, NC-15 was negative on PODs 0 and 182, and NC-16 was negative for C4d on POD 2.


  1. Top of page
  2. Abstract

We observed 2 different scenarios among transplant patients with a PC. A minority of our patients with a PPC (only 20%) maintained their PC and developed early AMR within the first weeks after transplantation. We presented 3 cases of rapidly and aggressively progressing AMR developing in the first weeks after transplantation in presensitized recipients with documented DSA levels and strong PCs. All those who developed evidence of graft dysfunction, the presence of DSAs, morphological tissue destruction consistent with a previous description of AMR, and positive C4d linear staining on the graft sinusoidal endothelium experienced transiently improved graft function with attempts to eliminate DSAs, but they subsequently developed a continuing severe course of rejection.

D-AMR patients required significant treatment in an attempt to reduce DSA levels. All patients displayed worsening graft function on standard immunosuppression and also with increased dosing of standard agents. One patient (D-2) treated for AMR died, likely because of the combination of a failing graft and sepsis. In liver transplant patients, the postoperative mortality rate remains approximately 10%, with bacterial infections being the leading cause of early mortality.26 Variables associated with bacteremia in liver recipients include Laennec's cirrhosis, diabetes, renal dysfunction, Acute Physiology and Chronic Health Evaluation II scores, rejection episodes, intensive care unit stay, and lower serum albumin levels.27 Patient D-2 had at least 4 of these factors. An additional factor could have been related to the escalation of immunosuppressive therapy. Another patient (D-1) chose not to undergo retransplantation; that patient subsequently suffered from exacerbation of hepatopulmonary syndrome due to a failing liver graft and died of pneumonia during supportive treatment. The third patient (D-3) successfully underwent retransplantation, and his second liver allograft followed the pattern seen in the ND-AMR group of patients with stable graft function 2 years after retransplantation.

Patient ND-1 sustained her PC for months with high levels of DSAs. A liver biopsy sample revealed strong and linear positive C4d staining in the patient's liver sinusoids. Because of a lack of clinical and histological evidence of liver injury, this patient was ND-AMR. The finding of positive C4d staining in a patient with minimal morphological or clinical graft injury may be similar to that described in kidney allografts in the Banff criteria.22, 23 On the basis of the presence of DSAs and positive sinusoidal staining for C4d, the concern was raised that AMR may develop over time, and the patient received a dose of rituximab.

It is obvious that in patients who undergo transplantation with a preformed alloantibody or who develop a de novo alloantibody after transplantation, a wide range of graft outcomes can be observed. These outcomes may range from devastating hyperacute rejection to early progressive graft dysfunction (possibly from acute AMR) to normal transplant allograft performance with or without positive IF staining for sinusoidal C4d. The last suggests a balanced immunological process similar to that described for graft accommodation after successful ABO-incompatible kidney transplantation.28

An early report that expanded our understanding of alloantibodies in liver transplantation relied on a pretransplant positive complement-dependent cytotoxic crossmatch (CDC) and evidence of liver allograft dysfunction with a posttransplant syndrome defined by a declining platelet count, declining total hemolytic complement activity, increasing levels of immune complexes, and PC persistence tested by CDC methodology.13 Currently, more than 40% of HLA laboratories have replaced the CDC crossmatch with flow cytometry crossmatching according to an American Society for Histocompatibility and Immunogenetics survey. Platelets, immunological complexes, and complement measurements cannot be applied to our study because of current therapies using antilymphocyte antibodies (Thymo and OKT3), PP, and/or IVIG. These treatments deplete platelets, complement, immune complexes, and clotting factors.29-34 Similarly, PP often requires replacement with fresh frozen plasma, and this invalidates the prothrombin time as a variable to follow. The measurement of DSAs via solid-phase technology and the addition of C4d IF testing of fresh frozen biopsies allow patients to be immunologically monitored during therapy and over long periods of time. In addition, the CDC method is a technique of highly variable sensitivity, and an NC does not confer a low risk of AMR.24 A positive CDC, regardless of titers of the measured cytotoxicity, is not a predictor of posttransplant AMR in liver allotransplantation, as indicated in the past.13 A recent study by Goh et al.35 describes poor outcomes for second liver transplants in recipients with high levels of a class I alloantibody, although there was no association detected with DSAs, and only CDC crossmatches were performed.

Complement is a part of innate immunological responses and becomes activated in a variety of immunological events, such as ACR and viral and autoimmune hepatitis36, 37; in addition to being produced by the liver, it is synthesized by human monocytes and macrophages.38, 39 This may explain the granular periportal C4d deposits observed in patients with ACR at the location of the leukocyte infiltrate in the periportal area but not on the sinusoidal endothelium, as seen in cases of isolated AMR.

In AMR, the main source of complement is likely the blood, and this results in C4d deposition at the sinusoidal endothelium. In liver preservation injury, C4d deposits were not identified on the sinusoidal endothelium.1 All C4d staining was performed on biopsy samples obtained before patients developed any signs of infection.

An earlier study found C1q and/or β1C/β1A components of complement and IgG/IgM; this was shown by IF of weak to moderate intensity deposited in sinusoidal and periportal veins of liver graft specimens collected from necropsy or graft hepatectomy samples (mostly on PODs 100-878).40 In the majority of the cases, the predominant finding was morphological features of chronic rejection with mononuclear cell infiltration in portal tracts by H&E. This by itself may affect the interpretation of complement deposits.39 The C4d molecule has the advantages of sensitivity and specificity over the C1q marker of complement deposition, and it has a longer half-life on tissues in comparison with C1q, IgM, or IgG.41 This earlier publication40 provides no information on the pretransplant sensitization status of the recipients, the pretransplant and posttransplant crossmatch results, or the DSA levels.

In this report, we confirm the findings of our previous publication1 on the importance of linear staining for C4d in liver sinusoids as one component in the diagnosis of AMR. The cases of isolated AMR showed strong linear C4d deposits in sinusoids according to IF microscopy of frozen tissue samples. Alternately, ACR case controls (patients without detectable alloantibodies) demonstrated weak granular C4d deposits in periportal areas only.1 This is consistent with previous reports.42 In a variety of publications, reports of C4d staining in terms of the location and pattern of deposits are conflicting (sinusoidal areas versus periportal areas), and the studies have been generally performed with stored specimens stained retrospectively on formalin-fixed and paraffin-embedded tissue sections.43, 44 We reported significant discrepancies between C4d staining performed on formalin-fixed, paraffin-embedded liver tissue and staining performed on frozen tissue samples with IF microscopy.45 This observation suggests that similarly to the evaluation of renal allograft biopsy samples, IF of frozen sections of liver tissue is the gold standard for the evaluation of C4d.46 Similar concerns regarding the application of immunohistochemistry of formalin-fixed and paraffin-embedded liver tissues have been raised in the past by others.12

Positive C4d IF staining of arteriolar internal elastic lamina is considered a normal intrinsic finding in the nephropathology literature.47 We also identified similar staining in the internal elastic lamina of hepatic arterioles in over half of our NC controls (Table 3). The Banff pathology conference of 2007 reiterated that focal C4d staining is still considered nondiagnostic on the basis of conflicting reports, mostly in the renal literature.22 In our evaluation and opinion, portal stroma staining in specimens collected from nontransplant patients and from NC recipients demonstrated a lack of specificity related to the presence of DSAs and AMR in liver allotransplantation. It appears that linear sinusoidal C4d staining by IF on frozen liver biopsy tissue is supportive of making the diagnosis of liver allograft AMR. Further studies with an applied uniform technique of C4d determination may help us to expand our experience with this marker in liver transplantation.

We have observed that patients maintaining PCs after transplantation are at risk of antibody-mediated liver rejection that is difficult to control or reverse. The number of patients with AMR may be minimal, but in our series, their liver grafts had disastrous outcomes. Because we cannot presently predict which PC recipients will develop AMR, we postulate that liver transplant recipients with PPCs should be closely monitored with a repeat crossmatch within 1 to 2 weeks after transplantation and again at 3 to 5 weeks. In our experience, patients who have a PC at 1 to 2 weeks have developed clinical AMR. These patients should be considered for early liver biopsy for pathological examination for AMR by C4d IF and standard light microscopy.

We believe that the spectrum of biological responses to a PC that we have observed explains the conflicting outcomes for transplant patients with PC in the literature. A persistent PC after liver transplantation can lead to early, severe AMR that may cause liver failure and death. Linear C4d staining in liver sinusoids along with graft dysfunction, morphological evidence for tissue injury, and the demonstration of DSAs seems to be pathognomonic for liver AMR. Patients with a positive pretransplant crossmatch should have the crossmatch repeated within the first 2 weeks after transplantation. Recipients with an unknown preoperative crossmatch who develop early cholestasis of unclear cause should be evaluated for AMR.


  1. Top of page
  2. Abstract