donor-specific anti–human leukocyte antigen DR antibody
idiopathic portal hypertension
mean fluorescence intensity
primary biliary cirrhosis
primary sclerosing cholangitis
Antibody-mediated rejection (AMR) in liver allografts is recognized as a possible cause of early and late allograft injury and a poor prognosis.[1-8] However, unlike acute cellular or chronic rejection, the diagnosis of AMR in liver allografts is often difficult to establish. One of the main reasons for this is the difficulty in interpreting complement component 4d (C4d) deposition, which is the most widely used marker of clinical AMR in renal, cardiac, and pancreatic transplantation.[8-11]
The specificity of C4d staining in liver allografts is controversial. Ali et al. and Lunz et al. correlated diffuse portal tract vascular endothelial C4d deposition with AMR. However, C4d positivity has also been reported for other medical conditions, such as acute cellular rejection (ACR),[1, 3-5, 12, 13] chronic rejection,[3, 5, 12, 13] ischemic injury,[1, 3, 12] hepatitis,[1, 3, 4, 14] and cholangitis.[1, 3, 4] Unfortunately, most of these previous studies performed C4d staining on nonconsecutive biopsy samples from unstable liver grafts. A more comprehensive study is required to understand the significance of C4d and its utility in AMR in combination with tests for alloantibodies.
In addition, sites of C4d deposition differ between and within these reports and include portal vessels,[1-6, 12, 14] portal stromata,[1, 2, 5] and sinusoids.[3-5, 12] The lack of agreement in staining patterns may also be related to the low specificity of C4d for AMR and may prevent clinicians and pathologists from using C4d for the routine histological diagnosis of liver allografts.
Kozlowski et al. recently suggested that strong linear staining in the sinusoid, rather than the portal tract, was a better marker for AMR, and they recommended the use of immunofluorescence on frozen sections. As they pointed out, immunoperoxidase staining is insensitive, and frozen sections may be a better tool for demonstrating C4d deposition. However, frozen sections are not suitable for conventional histological evaluations, and formalin-fixed, paraffin-embedded tissue is additionally required. Considering the rarity of clinical AMR in liver transplantation (LT), we suggest that establishing a method for evaluating C4d with immunoperoxidase alone may be practical.
Here we designed a nonselective, prospective study in which we performed C4d staining on all liver allograft biopsy samples obtained over the course of 4 consecutive months, and every clinically indicated biopsy was included in the study. The presence of anti–blood group (anti-A/B) antibodies or anti–human leukocyte antigen (anti-HLA) antibodies was evaluated during the same period. All patients were followed up for 2 years to clarify the significance of C4d in liver allografts. We adopted endothelial staining for this study, although we previously reported the stromal deposition of C4d as an ominous sign of ABO-incompatible (ABO-I) LT. The main reason for excluding stromal staining from this study is that only endothelial staining has been used as the standard for other solid organ transplants. Another reason is that stromal staining alone is often difficult to differentiate from the nonspecific staining seen in elastic fibers or necrotic tissue.[1, 17] When we picked up every portal stromal or endothelial stain, C4d staining was seen in various types of liver allograft injuries and did not show clinical significance. Because extensive C4d staining covers the endothelia of portal, sinusoidal, and perivenular areas,[1, 2] we now assume that endothelial staining alone is adequate for evaluating C4d.
PATIENTS AND METHODS
Study Population and Biopsy Samples
In a prospective and nonselective manner, regardless of indication, we studied all liver allograft biopsy samples obtained between July and October 2011 at Kyoto University Hospital. Patients who underwent LT outside Kyoto University Hospital were not included. Liver allograft biopsy was performed to determine allograft dysfunction or evaluate graft fibrosis when immunosuppression weaning was intended. If a patient underwent multiple biopsies during this period, the first biopsy sample that showed C4d positivity was selected for analysis. When all biopsy samples were negative for C4d, the first biopsy sample was selected. In each case, the biopsy specimen for analysis was classified as early (taken less than 30 days after transplantation) or late (taken 30 days or more after transplantation). All patients were followed until July 2013. Clinical and serological data were obtained from electronic patient charts. The institutional review board of Kyoto University approved this study.
The baseline immunosuppression protocol consisted of tacrolimus and oral prednisolone for both ABO-compatible or ABO-identical (ABO-C) patients and ABO-I patients. The lower limit of the target for whole blood tacrolimus levels was 10 to 15 ng/mL during the first 2 weeks, 10 ng/mL thereafter, and 5 to 8 ng/mL from the second month onward. Acute rejection was treated with a 3-day course of intravenous methylprednisolone bolus therapy (10 mg/kg). Mycophenolate mofetil was administered to patients who underwent refractory rejection or plasma cell hepatitis simulating autoimmune hepatitis. Select pediatric patients were weaned from immunosuppression according to the previously described protocol. All ABO-I patients underwent preoperative plasmapheresis or blood exchange in order to reduce anti-donor A/B antibodies to 1:8 or lower. In addition, patients who underwent ABO-I transplantation after 2006 received rituximab (an anti-CD20 monoclonal antibody) approximately 2 weeks before transplantation. Adult patients were given prostaglandin E1 and methylprednisolone via a portal vein or hepatic artery. Clinical AMR, which consisted of an elevation in postoperative anti-donor A/B antibody titers and graft dysfunction, was treated for approximately 5 days with plasmapheresis or intravenous immunoglobulin with steroid bolus therapy.
Liver allograft biopsy samples were processed for routine light microscopy. Biopsy specimens were fixed in 10% buffered formalin, sliced 3 μm thick, and stained with hematoxylin and eosin (H&E), Masson trichrome, and cytokeratin 7 (clone OV-TL 12/30, Dako, Glostrup, Denmark; 1:200 dilution).
ACR and chronic rejection were diagnosed according to the Banff criteria.[20, 21] AMR was diagnosed according to the criteria used for other solid organ transplants: (1) clinical evidence of graft dysfunction, (2) histological evidence of graft injury, (3) immunopathological evidence of antibody action (C4d deposition), and (4) serological evidence of anti-HLA or anti-donor antibodies at the time of biopsy. A combination of periportal edema, hemorrhage, and neutrophilic infiltration was regarded as an indicator for AMR in ABO-I patients.[8, 23] Allograft fibrosis was staged according to the METAVIR scoring system.
C4d Immunohistochemical Staining
A rabbit polyclonal anti-human C4d antibody (Bl-RC4D, Biomedia; 1:50 dilution) was used to detect C4d. Staining was performed on a Ventana Benchmark Ultra autostainer. Sections were treated with protease (Ventana; 0.5 U/mL) at 37°C for 20 minutes for antigen retrieval. C4d immunostaining for formalin-fixed, paraffin-embedded tissue was first available at our laboratory in August 2003, but it was applied to only select cases and was not used routinely before this study.
Staining was recorded as diffuse when linear C4d deposition in the portal tract vascular endothelium was seen in 50% or more of the portal tracts. Staining of fewer than 50% of the portal tracts was considered focal. We also evaluated the intensity of staining, which was recorded as strong when linear C4d deposition was seen with low-power magnification (4× objective lens) and as weak when staining was confirmed only at a higher magnification. Completely negative staining (score = 0) and focally weak staining (score = 1) were considered negative and equivocal, respectively. Diffuse or strong staining (score = 2) and diffuse and strong staining (score = 3) were considered positive for the statistical analysis. Staining in hepatocytes, portal stromata, and elastic fibers was recorded but was not included in the statistical analysis. All stained slides were interpreted by M.F. and H.H. without clinical data.
Assays for Alloantibodies
The lymphocyte cross-match test was conducted only before transplantation. After LT, the anti-HLA antibody titer was analyzed with Luminex multiplex technology at the time of biopsy. The specificity of positive tests was determined with the LABScreen single-antigen test (LABScreen mixed and single-antigen tests, One Lambda, Canoga Park, CA), and the results were displayed as mean fluorescence intensities (MFIs). An MFI > 5000 was regarded as positive. The anti-HLA antibody was then compared with the patient's HLA type to determine whether it was a donor-specific anti–human leukocyte antigen antibody (DSA) or a non-DSA.
In ABO-I cases, serum levels of anti-A/B antibodies were evaluated before and after LT with the microhemagglutination assay. This test was conducted at least 3 times per week during the first postoperative month. A postoperative anti-donor blood group immunoglobulin M titer of 1:32 or more was defined as an elevated titer.
Associations between categorical variables were assessed with Fisher's exact test. Descriptive statistical methods (means, medians, standard deviations, and ranges) as well as the Mann-Whitney U test were used to assess the distributions of variables. For all analyses, a P value < 0.05 was regarded as significant.
In all, 219 biopsy samples were obtained from 163 patients (range = 1-9 per patient) during the study period. After the exclusion of 20 ABO-C patients whose Luminex assays for anti-HLA antibodies were not available (N/A) at the time of index biopsy, 143 patients with a total of 194 biopsy samples were enrolled in this study. Seven ABO-I patients who underwent isoagglutinin tests but not Luminex assays were not eliminated.
The demographics of the patients are summarized in Table 1. Most patients (98%) underwent living donor LT. The most common indications for transplantation in the pediatric and adult groups were biliary atresia (BA) and chronic hepatitis C (HepC), respectively. In the ABO-C group, there were 114 patients: the percentage of children (age < 18 years) was higher (74% versus 38%), and for most (91%), the index biopsy was performed 30 days or more after transplantation. In the ABO-I group, there were 29 patients, and ACR, C4d positivity, and graft loss were more commonly seen in comparison with the ABO-C group. All patients were lymphocyte cross-match–negative before transplantation. No significant difference was observed in the percentage of positivity for anti-DSA antibodies between the ABO-C group and the ABO-I group. We also checked the data with an MFI cutoff point of 1000, and there was no difference between the 2 groups (data not shown). The distribution of DSAs by class among the patients was as follows: class I, 1; class II, 36; and classes I and II, 3 . Among the 39 patients with anti–class II antibodies, antibodies against DR loci were most commonly observed (n = 27 or 69%). Among the 96 DSA–negative patients, 22 showed non-DSAs (MFI > 1000), 7 showed weak class II antibodies (MFI > 1000 but ≤ 5000 against the donor DR locus), 2 showed weak class I antibodies, and 65 were completely negative for anti-HLA antibodies.
Table 1. Comparison of ABO-C Patients and ABO-I Patients
ABO-C Patients (n = 114)
ABO-I Patients (n = 29)
The data are presented as medians and ranges.
The C4d scores for the endothelium of portal areas were determined as follows: (0) completely negative staining, (1) focal and weak staining, (2) diffuse or strong staining, and (3) diffuse and strong staining.
Three ABO-C patients and 6 ABO-I patients died during the follow-up period, and none of them showed positivity for anti-HLA antibodies or high anti-A/B antibody titers. For 2 ABO-I patients, Luminex assays were not performed before death. All the ABO-C patients were negative for C4d, whereas 5 of the 6 ABO-I patients (83%) showed C4d positivity. Four patients died of a severe bacterial or fungal infection within 6 months of LT. The other 5 patients died of severe ACR (7 months after LT), graft-versus-host disease (14 months after LT), fibrosing cholestatic (HepC; 15 months after LT), ischemic cholangiopathy after rupture of the hepatic artery (6 years after LT), or cirrhosis due to de novo autoimmune hepatitis (14 years after LT), respectively.
Characteristics of C4d-Positive Cases in ABO-C Transplantation
Table 2 lists the clinical and histological characteristics of 20 patients exhibiting C4d positivity on index biopsy. According to early biopsy samples for the ABO-C group, only 1 of 10 patients was positive for C4d (case C1), and a statistical analysis was not suitable for this subgroup (Table 3). A previous biopsy sample from C1, which was taken on postoperative day (POD) 7 and showed a moderate degree of ACR, was also C4d-positive but was outside this study period.
Table 2. Characteristics of Patients Showing C4d Positivity in the Endothelium
In late biopsy samples from the ABO-C group, C4d immunoreactivity was significantly correlated with graft bridging fibrosis (P = 0.01) but not with ACR histology, levels of serum transaminases, or total bilirubin (Table 3). Although positivity for anti-DSA antibodies was not statistically associated with C4d positivity, the presence of DSAs against DR loci was correlated with the C4d status (P = 0.04). The inclusion of the anti–HLA-DQ antibody status made the difference statistically insignificant (data not shown). When late biopsy samples were divided in terms of donor-specific anti–human leukocyte antigen DR antibodies (HLA-DR DSAs), the presence of HLA-DR DSAs was significantly associated with fibrosis, ACR, and C4d scores but not with levels of serum transaminase or total bilirubin (Table 4).
Table 4. Correlations of HLA-DR DSAs and Clinicopathological Parameters in Late Biopsy Samples From ABO-C Patients
MFI > 5000 (n = 25)
MFI ≤ 5000 (n = 79)
The data are presented as means and standard deviations.
Cases with C4d-positive late biopsy samples included heterogeneous histological findings with various possible causes of fibrosis (C2-C5 in Table 2); C2 and C3 were pediatric cases whose protocol biopsy samples showed minimal or no inflammatory cell infiltration, and the C4d positivity was thought to be related to suboptimal immunosuppression.
Case C4 was a patient whose recurrent HCV had been treated with interferon since 7 months after LT with the diagnosis of stage 1 fibrosis. Although a sustained virological response was achieved, a biopsy sample taken 5 years after LT revealed progression of his fibrosis and focal ductopenia (Fig. 1A,B). This patient was found to have a low titer of anti-nuclear antibodies, but the histological findings were different from those for autoimmune hepatitis and were compatible with chronic cholangiopathy (Fig. 1C). There was a history of a biliary anastomotic stricture 2 years after LT, and the patient underwent a successful removal of biliary casts. Diffuse C4d staining was noted in fibrous portal tracts (Fig. 1D), and C4d positivity persisted in a biopsy sample taken a year after this study period. Another patient with a history of recurrent HCV (case C5) was DSA-negative at the time of index biopsy with interferon therapy. When a follow-up biopsy was performed after the cessation of unsuccessful interferon therapy, the C4d findings were negative (Table 2).
Although none of the patients undergoing ABO-C LT in this study period were diagnosed with clinical AMR, 1 patient was revealed to have persistent graft dysfunction along with persistent DSAs and a history of sporadic C4d staining. The patient was not included in Table 2. Before transplantation, the lymphocyte cross-match test was negative, and the Luminex test was N/A. Three allograft biopsies within the first 3 months after transplantation showed ACR, and C4d staining was negative each time. Despite the long-term use of triple immunosuppression (tacrolimus, prednisolone, and mycophenolate mofetil), graft dysfunction persisted, and the histological diagnosis after 6 months was mild ACR with perivenular hemorrhage (Fig. 2A). Diffuse endothelial C4d staining with some stromal staining was seen on biopsy samples taken on PODs 185, 192, and 227 (Fig. 2B). The Luminex test revealed DSAs on POD 229 (B59, 3932; DR4, 15,840; DR53; 8061; DQ4, 4747). During this study period (POD 524), portal inflammation was mild (Fig. 2C), and C4d staining was faint and considered negative (Fig. 2D). DSA results remained positive (B59, 3434; DR4, 12,318; DR53, 2444), and portal and perivenular fibrosis progressed (Fig. 2E). Serum bilirubin levels remained at 2 to 3 mg/dL. On the last follow-up biopsy sample taken on POD 986, DSA results remained positive (B59, 4509; DR4, 6458; DR53, 23,557; DQ4, 23,738) with persistent fibrosis and a ductular reaction. Bile duct loss was not observed. C4d endothelial staining returned (Fig. 2F).
Characteristics of C4d-Positive Cases in ABO-I Transplantation
In both early and late biopsy samples, the C4d status for ABO-I LT patients was not statistically associated with any clinical parameters possibly related to rejection (Table 5). The majority of C4d-positive patients did not show postoperative elevations in anti-donor A/B antibody titers despite C4d endothelial staining (cases I1-15; Table 2). Only 3 patients (I1, I8, and I12) showed anti-A/B antibody titer elevations, and they were the only patients who fulfilled the criteria for AMR: (1) detectable anti-donor antibodies (1:32 or more anti-A/B antibodies with or without the presence of an anti-HLA antibody), (2) C4d in the graft endothelium, (3) graft pathology, and (4) graft dysfunction. These 3 patients showed typical ABO-I–associated injuries, which were characterized by portal edema and hemorrhage with foci of necrosis (Fig. 3A). Sinusoidal C4d staining was also observed in case I8 (Fig. 3B). All ABO-I AMR cases responded well to steroid pulse therapy with or without plasmapheresis and immunoglobulin bolus administration. The level of isoagglutinin decreased to 1:4 or lower after therapy for AMR. Follow-up biopsy samples showed diffuse C4d positivity in case I8, equivocal (score 1) staining in case I1, and complete negativity in case I12 (Table 2) 59, 390, and 169 days after index biopsies, respectively.
Table 5. Correlation of C4d Positivity and Clinicopathological Parameters in ABO-I LT
Early Biopsy (<30 Days After Transplantation)
Late Biopsy (≥30 Days After Transplantation)
C4d-Positive (n = 5)
C4d-Negative (n = 2)
C4d-Positive (n = 10)
C4d-Negative (n = 12)
The data are presented as means and standard deviations.
All C4d staining for ABO-I LT patients tended to fade in the follow-up biopsy samples. Only in 3 of the last 11 follow-up biopsy samples did the C4d scores remain the same as those of the index biopsy samples (cases I1, I6, and I15; Table 2).
This study shows that C4d positivity without an elevation in anti-donor A/B antibodies is not uncommon among patients with ABO-I LT. Before the use of rituximab, we observed that postoperative isoagglutinin titer elevations were often associated with fatal AMR, which was characterized by periportal edema, necrosis, and hemorrhage.[2, 23] C4d deposition was commonly seen in portal stromata as well as the endothelium. In contrast, all ABO-I transplant recipients in this study underwent planned preoperative intravenous rituximab administration as well as plasmapheresis or blood exchange. As a result, most of the C4d-positive ABO-I cases had low anti-A/B antibody titers at the time of biopsy and did not show histological evidence of critical graft injury. This is somewhat similar to Haas et al.'s findings in ABO-I kidney allografts. This result may also be explained by a consideration of the liver's ability to absorb, eliminate, and neutralize antibodies. Mild alloantibody reactions may cause C4d deposition but not significant allograft injury.[8, 27] Another possibility is the presence of the accommodation phenomenon. In ABO-I renal allografts, graft resistance to the acute pathological effects of graft-specific antibodies even after the rebound of antibody concentrations has been called accommodation. However, in our series, cases with postoperative elevations in anti-A/B antibody titers were associated with periportal changes that were compatible with acute antibody-mediated allograft injury accompanied by the focal or diffuse deposition of C4d. This suggests that postoperative titer monitoring may be practical for predicting acute AMR in patients undergoing ABO-I transplantation and that the routine application of C4d immunostaining in ABO-I LT may not be necessary for detecting acute AMR.
Diffuse or strong C4d staining was uncommon in ABO-C cases, and none of the C4d-positive cases during the study period were associated with typical severe allograft rejection. We previously reported that lymphocyte cross-match–positive transplantation without preventive conditioning against AMR could result in clinical AMR.[1, 25] In that report, lymphocyte cross-match–positive cases often showed diffuse C4d positivity, and common histological findings were ACR, neutrophilic cholangitis/cholangiolitis, and hepatocanalicular cholestasis. After encountering some fatal clinical AMR cases, we tried to avoid lymphocyte cross-match–positive transplantation. Therefore, patients in this study were all negative for lymphocytic cross-match tests before LT; C4d positivity was not associated with severe inflammation or cholestasis, which could suggest acute AMR after ABO-C LT. We suggest that avoiding cross-match–positive LT reduced critical AMR, but C4d-positive cases may still be observed without severe graft damage.
As in studies of renal allografts, an association of DSAs and chronic rejection has been recognized in some studies of LT.[5, 13] We reported that anti–class II DSAs were related to late graft fibrosis and C4d positivity. This study also proved that HLA-DR DSAs were associated with late-onset acute rejection, graft fibrosis, and C4d deposition. Although the previous study focused on pediatric cases and excluded fibrosis with apparent causes such as steatohepatitis, this study included all biopsy samples from adult and pediatric patients whose fibrosis could be attributable to nonrejection episodes. It is notable that 2 adult patients who were treated with interferon for recurrent HCV were included among the C4d-positive cases. Because HepC itself is associated with graft fibrosis, it seems difficult to determine whether C4d has a role in graft fibrosis. Interferon therapy alone may be related to C4d positivity. In 1 of the 2 patients, however, a progression of fibrosis was observed even after a sustained viral response and the successful treatment of a biliary stricture. Diffuse C4d positivity and persistent anti–class II (DR locus) DSAs might be related to progressive fibrosis and bile duct loss. In addition, a pediatric case in whom C4d positivity was found before this study was also associated with progressive fibrosis, which was a clue for proving DSAs. These findings suggest that C4d can be a tool for detecting possible DSA-related fibrosis; the causes of fibrosis can be multifactorial, especially among adults, who may experience a recurrence of their original disease and have a positive DSA status at the same time. Because C4d positivity was rare and was not associated with graft loss or severe graft dysfunction, C4d immunohistochemistry seems to be useful for the evaluation of late allograft biopsy samples only in limited situations, such as immunosuppression weaning and unusual allograft fibrosis. However, C4d staining is inexpensive and can be easily evaluated with conventional biopsy samples, and it would be more practical than applying HLA assays in all cases after LT. The exact prognostic significance and contribution to the optimization of immunosuppressants need to be determined in future studies.
Our study has several limitations for the analysis of DSAs. Preoperative data from HLA assays other than lymphocyte cross-match tests were N/A in most cases. Postoperative HLA assays were not performed during a fixed period of time after LT. Although the negativity of preoperative lymphocyte cross-match tests suggests that most DSAs found in late biopsy samples were associated with de novo DSAs, definitive data are lacking in this study. Because the presence of DSAs did not correlate with the levels of serum transaminases or total bilirubin, further study of alloantibodies and autoantibodies is also required in order to clarify the presence of chronic AMR of the liver; assays for immunoglobulin subclass or complement fixation might be more important than the simple quantification of those antibodies.
In conclusion, our study is the first to compare the prevalence of C4d positivity in ABO-C and ABO-I liver allografts through the application of C4d immunohistochemistry to routine anatomic pathology practice. In ABO-C LT, diffuse or strong endothelial C4d positivity is uncommon and may be associated with graft fibrosis and the presence of HLA-DR DSAs. In ABO-I LT, C4d positivity is common with or without elevations in postoperative anti-A/B antibody titers and has little value in detecting acute AMR.