|Home | About | Journals | Submit | Contact Us | Français|
The role of humoral alloreactivity in ABO-compatible liver transplantation remains unclear. To understand the significance of donor-specific HLA alloantibodies (DSA) in liver rejection, we applied the currently used strategy for detection of antibody-mediated rejection of other solid allografts. For this purpose we reviewed the data on 43 recipients of ABO identical/compatible donor livers who had indication liver biopsy stained for complement element C4d and contemporaneous circulating DSA determination. Seventeen (40%) patients had significant circulating DSA in association with diffuse portal C4d deposition (DSA+/diffuse C4d+). These DSA+/diffuse C4d+ subjects had higher frequency of acute cellular rejection (ACR) 15/17 versus 13/26 (88% vs. 50%), p = 0.02, and steroid resistant rejection 7/17 versus 5/26 (41% vs. 19%), p = 0.03. Based on detection of the combination DSA+/diffuse C4d+, 53.6% of cases of ACR had evidence of concurrent humoral alloreactivity. Six of the 10 patients with ductopenic rejection had circulating DSA and diffuse portal C4d, three of whom (2 early and 1 late posttransplantation) developed unrelenting cholestasis, necessitating specific antibody-depleting therapy to salvage the allografts. Thus, in ABO-compatible liver transplantation humoral alloreactivity mediated by antibodies against donor HLA molecules appears to be frequently intertwined with cellular mechanisms of rejection, and to play a role in ductopenia development.
The distinct role of humoral mechanisms (1) is well established for kidney (2–4) and heart (5) allograft rejection and convincing evidence is emerging to support its role also in pancreas (6) and lung (7) transplantation. Although experimental evidence of alloantibody-mediated rejection of the liver was provided more than 20 years ago (8), supported by early clinical observations (9–11), humoral rejection (HR) in clinical transplantation of ABO identical/compatible livers has been viewed as insignificant, the liver being considered an immunologically privileged organ resistant to HLA alloantibody (12,13).
Recent clinical evidence has led to reconsideration of these views. Although the donor-recipient compatibility has been shown not to have an impact on the global liver allograft survival, the presence at transplant of antibodies against donor HLA as determined by Luminex multibead assay or complement-dependent cytotoxicity crossmatch was found to be associated with shorter 1- and 5-year graft survival, to correlate with allograft rejection (14) and to predict a lower allograft survival after retransplantation (15), confirming earlier observations also based on preoperative detection of lymphocytotoxic antibodies (16,17). More recently, on the basis of immunohistochemical detection of the complement split product C4d, viewed as a reliable histological footprint of antibody mediated rejection (AMR) (18), a strong association was inferred regarding the detection of C4d in acute cellular rejection (ACR) of the liver allograft, C4d being recommended as a potential discriminator of rejection from recurrent hepatitis C (19). Another study corroborating the presence of C4d staining with the preoperative detection of donor-reactive lymphocyte antibodies by crossmatch found that diffuse C4d deposition in crossmatch-positive liver recipients was associated with poor outcome (20). So far however, only a few individual case reports, including from our institution (21–25), have based the diagnosis of humoral alloreactivity of the liver on modern methods currently in clinical use to diagnose AMR of other allografts such as kidney, pancreas and heart, namely tissue C4d immunolabeling in conjunction with concurrent donor-specific HLA alloantibodies (DSA) detection. In this study we corroborated the C4d deposition in the liver and the contemporaneous detection of DSA with cellular and ductopenic rejection, and the therapeutic intervention. We believe this is the first study to systematically analyze in a larger number of patients the significance of DSA as determined by single antigen beads, solid phase flow cytometry (Luminex) in the rejection of ABO identical/compatible liver allografts and to provide what was considered in a recent editorial (13) the link missing in the previous humoral alloreactivity studies based on C4d detection in the liver tissue, namely the correlation of C4d immunolabeling with circulating DSA.
A search of our electronic Transplant Data System revealed that 740 clinically indicated liver biopsies to evaluate liver dysfunction were obtained from 364 allografts in 346 patients between April 1, 2006 and February 28, 2009. Of the 740 biopsies, 187 showed ACR (mild: 111, moderate: 60, severe: 16) in 153 patients, and 40 showed ductopenic rejection in 25 patients. Of this cohort of 346 patients, 43 nonconsecutive cases had information on both C4d immunolabeling of the liver tissue and donor-specific HLA antibodies determination by Luminex. The interval between the biopsy and DSA determination was under 2 weeks; in 36 cases (85%) the DSA being determined within 3 days of the biopsy. The decision to determine the DSA and C4d was made by the individual managing physician. Our study, approved by the Health Sciences Institutional Review Board, is a retrospective review of the clinical and pathological data of these 43 patients.
ACR of the liver allograft was diagnosed by Banff criteria. Ductopenia was defined as loss of the native bile ducts in at least 50% of the portal tracts.
Histological evidence of acute cholangiolitis and biliary tract obstruction in the absence of radiographic evidence of large duct obstruction (LDO) has been considered a potential characteristic morphological feature of AMR (9,17,21). Therefore histological findings concerning for biliary obstruction including ductular reaction, cholestasis or frank LDO were reported in conjunction with the imaging studies. The ductular reaction was defined as the presence of increased number of bile ductules accompanied by polymorphonuclears (cholangiolitis) and LDO as a constellation of ductular reaction, canalicular cholestasis, feathery degeneration and portal edema (26).
Paraffin embedded livertissue underwent C4d immunolabeling using a 1:50 dilution of rabbit polyclonal C4d antibody (American Research Products, Belmont, MA, USA). For antigen recovery of organs, the formalin-fixed tissue went through deparaffinization and mild heat-induced recovery of organs with CC1. Kidney tissue with HR was used as positive control.
Semiquantitative evaluation of C4d immunolabeling was performed by assessing staining of portal capillaries and stroma (denoted together as portal-C4d), portal vein endothelium, sinusoidal endothelium, terminal hepatic venular endothelium and hepatic artery. Diffuse staining was defined as C4d deposition in portal capillaries/stroma of more than 50% of the tracts. Portal-C4d immunolabeling of fewer than 50% of tracts was considered focal. Similarly, staining of the sinusoids was graded as diffuse if greater than 50% of the sinusoidal endothelial bed was stained, and a lesser degree of staining was recorded as focal.
Serum specimens were analyzed for HLA antibodies by LabScreen® Single Antigen Class I and Single Antigen Class II beads (SAB) Luminex kits from One Lambda, Inc. (OLI, Canoga Park, CA, USA), following the method described in the package insert. The resultant LXFC output files were analyzed with HLA Visual® analysis software (OLI) and normalized mean fluorescence intensity (MFI) values to donor mismatched HLAs were reported. The specificities for both class I and II HLA antibodies were considered positive for an MFI > 500, a value above the cut-off for our laboratory chosen based on recently published data for HR of kidney and pancreas (6,27).
Steroid-resistant acute rejection was defined as biopsy-proven acute rejection not responding to steroid pulses, in general dexamethasone 20–100 mg intravenously for 3 days with subsequent transition to oral prednisone and steroid taper, along with enforcement of maintenance immunosuppression. Response was defined clinically as sustained improvement in the liver biochemical tests.
Continuous variables were summarized by reporting means ± SD and compared between groups using Wilcoxon rank sums tests. Categorical variables were summarized by reporting percentages and compared between groups using a Fisher's exact test. p-Values < 0.05 were considered significant. All analyses were performed using SAS statistical software version 9.1 (SAS Institute, Inc., Cary, NC, USA).
The characteristics of the 43 patients in this study are presented in Tables 1 and and2.2. There was no significant demographic or clinical difference between the DSA positive and negative patient. The donor-recipient ABO was identical in 42 subjects, and compatible in one (case 10, donor ABO: O, recipient ABO: B). Seventeen patients had indication liver biopsy in the early (<90 days) postoperative period (21 ± 20 days), and 26 patients beyond 90 days posttransplantation (1049 ± 1049 days).
Twenty-eight patients had histological evidence of ACR, which was mild in 17 (61%), moderate in 9 (32%) and severe in 2 (7%) patients. Ten patients had evidence of ductopenic rejection, of which six in association with ACR. In two cases the biopsies were indeterminate for rejection and no additional antirejection treatment was rendered.
Histological findings concerning for biliary obstruction but with negative radiographic studies, a characteristic of AMR, were found in 6 DSA-positive cases with portal C4d deposition and ACR (Table 3).
Twenty-seven patients had circulating DSA with MFI > 500. The MFI for class I DSA was 2005 ± 3027 (range 546– 7417), and for class II 4143 ± 3611 (range 630–14 808). The distribution of HLA antibodies by class among patients was as follows: 22 class II, 4 class I and II and 1 class I only (Table 3).
The anatomic pattern of C4d deposition is summarized in Table 3 and illustrated in the figure. There was no correlation between the extent of C4d deposition and the severity grade of ACR. The C4d immunolabeling of the portal capillaries and portal stroma was present in 34 patients, (diffuse in 23 and focal in 11). C4d staining of the endothelium of the portal vein was seen in 24 cases, 23 of which in associated with portal C4d deposition. The C4d immunolabeling of the sinusoidal endothelium was present in a diffuse pattern in seven, and in a focal pattern in five cases, 92% (11/12) having concomitant portal C4d deposition. Overall, 74% of the diffuse portal, 36% of focal portal and 58% of sinusoidal C4d staining cases had circulating DSA. Conversely, the DSA positive cases had C4d deposition as follows: diffuse portal in 63%, focal portal in 18.5%, sinusoidal in 26% and none in 18.5%.
The association of C4d patterns, overall or only in combination with DSA detection, with ACR is presented in Table 4. A statistically significant association was found between diffuse portal C4d deposition and ACR. Therefore, the detection of DSA positivity and diffuse portal C4d (DSA+/diffuse C4d+) deposition in combination with histological rejection was taken as evidence of complement-dependent alloantibody-mediated rejection specifically triggered by antidonor HLA alloantibodies. Lesser degree of portal C4d deposition in association with DSA was considered indeterminate for humoral alloreactivity.
Based on detection of the combination DSA+/diffuse C4d+, 53.6% of cases of ACR had also evidence of concurrent HR mediated by donor-specific HLA antibody. The DSA+/diffuse C4d+ subjects had a higher frequency of ACR 15/17 versus 13/26 (88% vs. 50%), p = 0.02, and of steroid resistant rejection 7/17 versus 5/26 (41% vs. 19%), p = 0.03.
Two patients had DSA+/diffuse C4d+ but lacked histological evidence of ongoing ACR, raising the possibility of smoldering humoral alloreactivity. Both had received pulse steroids in the preceding 3 weeks, patient no. 16 for ACR of the liver, and patient no. 17 for kidney allograft rejection. An additional confounder was the presence of biliary strictures and possibly subclinical cholangitis that may have contributed to complement activation and C4d formation.
Six patients had C4d deposition in a diffuse portal pattern unaccompanied by detectable DSA, four of which with ACR. Four of these six patients (nos. 28, 30, 31 and 32), with lone diffuse portal C4d also had biliary strictures and cholangitis.
Ten patients had evidence of ductopenia due to chronic rejection. All had patent hepatic arteries. Nine of the 10 cases had portal C4d deposition, seven diffusely and two focally. Seven patients had circulating DSA, six of which in association with diffuse portal C4d deposition. Among the 28 patients with ACR, six had evidence of concomitant ductopenic rejection. In this subgroup ductopenia was present only in cases with the combination DSA+/diffuse C4d+, 6/15 versus 0/13, p = 0.01.
In three patients ductopenia developed while on adequate maintenance immunosuppression. Patients no. 4 and 14 developed bile duct loss very early after transplant in conjunction with ACR. In one patient antithymocyte globulin (ATG) and in the other basiliximab and ATG were used with histological resolution of cellular infiltrate. Both however continued to have unrelenting cholestasis and bile duct injury, and required antibody-depleting therapy with plasmapheresis, intravenous immunoglobulin (IVIG), rituximab (case 14 only) and splenectomy, to control cholestasis and salvage the liver allografts. Case no. 11, who presented while on adequate maintenance immunosuppression in a somewhat similar manner but late posttransplant, also showed a significant improvement in cholestasis after treatment with IVIG and plasmapheresis. No follow-up biopsy exists in this patient who succumbed to sudden death while at home. The other seven patients had ductopenic rejection attributable to inadequate immunosuppression due to recurrent noncompliance (patients no. 7, 8 and 37), or intentional reduction of calcineurin inhibitors (CNI) necessary to control hepatitis C (no. 26 and 32) and CNI side effects (no. 6 and 40).
Follow-up biopsy with C4d staining was available for five cases (Table 5). Case no. 14 had several biopsies showing ongoing bile duct damage, ductopenia as well as persistent diffuse portal C4d and intermittently sinusoidal deposition, in spite of decreased serum DSA after treatment with PP and IVIG and later rituximab treatment. Only after splenectomy did the ductopenia resolve and C4d deposition markedly decreased (very faint, focal only staining in the last biopsy).
It is currently accepted that tissue deposition of C4d, the final split product resulting from the proteolytic cleavage of complement element C4, is a marker of humoral alloreactivity (18). In order to investigate the role of HLA antibody in complement-dependent HR of the liver allograft in ABO identical/compatible transplantation, we corroborated the tissue deposition of C4d with the concurrent detection of circulating DSA, a strategy currently in clinical use to diagnose AMRin other solid organ allografts.To our knowledge, except for a few case reports, no study has correlated the DSA detection by the sensitive and specific method of Luminex with C4d deposition in liver allografts undergoing acute and chronic rejection. Our subjects underwent clinically indicated liver biopsy to evaluate the cause of abnormal liver tests both in the early (<90 days) as well as the late posttransplantation period, and the study biopsies covered the severity spectrum of ACR, ductopenic rejection or no rejection, being therefore representative for the rejection types seen in our practice.
Methodologically, we performed for practical reasons immunostaining for C4d on paraffin embedded tissues, although some raised the possibility that this method may have lower sensitivity than staining of frozen tissue (28,29).
No consensus has been developed regarding the C4d pattern of significance for detection of HR in liver allografts. Most studies (Table 6) have reported C4d deposition in portal tracts, noticing immunolabeling of vascular structures (including endothelium of portal capillaries, veins, sinusoids, central veins and hepatic arteries), and portal stroma (20,29–31). In our study the most common site of C4d deposition was the portal capillaries and nearby stroma, presumably due to leakage of C4d from the damaged endothelium, and covalent binding via the exposed thioester group to the closest protein or carbohydrate in the matrix near the site of complement activation and C4d formation (32).
In our study, a heterogeneous C4d immunolabeling pattern was observed, and the diffuse portal C4d deposition had a statistically significant association with ACR. This is in line with other studies which have shown that the extensive, diffuse portal C4d deposition had clinical significance in terms of rejection and graft survival (20,31), and analogous with demonstration of humoral alloreactivity seen in other allografts, where the diffuse C4d deposition and DSA positivity is required for the diagnosis of HR, with yet unclear clinical significance of lesser degree of C4d deposition (4,33,34). We therefore considered the detection of circulating DSA with diffuse portal C4d deposition in combination with histological rejection an indicator of complement-dependent humoral alloreactivity specifically mediated by antibodies against donor HLA molecules.
Thirty percent of the cases falling in this category had also histological findings potentially characteristic of AMR, namely histological evidence of biliary outflow obstruction without radiological obstruction of the large bile ducts.
Sinusoidal endothelial C4d immunolabeling has been reported mostly in association with portal deposition (Table 6), with one group remarking on its absence (19,32). However in a well-documented case of pure HLA-AMR a predominantly diffuse sinusoidal C4d deposition was observed by immunofluorescence using monoclonal antibodies (21). We observed sinusoidal C4d immunolabeling almost exclusively in association with portal C4d staining. Hence, we could not determine the significance of isolated sinusoidal C4d deposition. Based on our findings, sinusoidal C4d staining seems to represent a more widespread C4d deposition extending beyond the portal tracts downstream into the sinusoids. As such it may add to the specificity of C4d staining. Had we taken into account only the cases with sinusoidal C4d deposition, we would have missed several cases of AMR, including one of severe acute ductopenia with initially only portal C4d deposition (case 14).
A great variability (8–100%) in C4d deposition in ACR of ABO compatible liver allografts has been reported (Table 6). Based on the presence of both diffuse portal C4d deposition and detection of circulating DSA we found that the frequency of complement-dependent humoral alloreactivity to donor HLA in ACR was 53.6%. Overall diffuse portal C4d irrespective of DSA status was present in 67.8% of ACR cases, close to figures reported by other of 67.7% (19) and 69.2% (35). While it is possible that the C4d deposition occurred in HLA-antibody negative patients as a result of complement activation by other types of donor-specific alloantibodies (36) it should also be noted that three of the four cases of lone diffuse C4d positivity associated with ACR had biliary obstruction and cholangitis. In these cases the C4d might have resulted from complement activation by infection via the lectin or C-reactive pathways (37,38). The detection of C4d should not however be invariably attributed to cholangitis when this infection is present. For example, case no. 27 had no C4 deposition in spite of having ischemic biliary strictures and severe bacterial cholangitis. Future prospective studies will allow consideration of other circumstances that might complicate C4d interpretation in liver allografts in the presence of inflammatory liver diseases shown to result in C4d deposition even in native livers (39,40).
The portal C4d deposition is pathogenically concordant both with the localization of inflammatory infiltrate and the overexpression of HLA antigens in the structures in this region during ACR. Indeed in acute rejection class I antigens are increased, and class II antigens, especially DR, DP but also DQ, become overexpressed on endothelial cells, and bile ducts (41,42). In our study antibodies against both HLA classes appeared involved in humoral alloreactivity, in conformity with other recent studies (14,31 and Table 6).
Earlier studies have postulated a role for lymphocytotoxic antibodies against donor HLA in the pathogenesis of ductopenia (9,43,44). Recent studies based on retrospective tissue immunolabeling demonstrated C4d deposition in chronic ductopenic rejection in 17% (45) to 100% (29,46) of cases. Seventy percent of the patients with ductopenia in our study had DSA, and 60% of ductopenia cases had both circulating DSA in association with diffuse portal C4d deposition, supporting a role for AMR in the pathogenesis of interlobular bile duct injury and loss. The resolution of cholestasis and ductopenia in association with reduction of C4d deposition only after decrease in circulating DSA with aggressive therapy specifically directed toward antibody removal further supports a role for DSA in the pathogenesis of ductopenia. Of particular importance is the risk of development of acute ductopenia as seen in this study, an operational distinction previously defined as an irreversible, rejection-related condition of acute vanishing bile duct syndrome occurring within 100 days after liver transplantation (47). Recognition of the role of HLA-AMR in the pathogenesis of this rare but serious entity is essential in order to salvage the allograft by timely initiation of specific antibody-depleting therapy.
Corroborating our findings of significant portal C4d deposition in DSA positive cases with the elegant demonstration by morphometric studies of portal microvasculature destruction preceding the bile duct loss in the process of rejection (48–50) we subscribe to the previously proposed mechanism in which the biliary tree may behave similarly to the kidney tubules because of their arterial only blood supply and conventional arteriolar and capillary plexus, the injury of which triggered by HLA antibodies could account for ductopenia (9). Therefore the following chain of events seems to occur: formation of DSA-HLA complex on endothelial cells of the portal tract microvasculature triggers complement activation (evidenced by C4d deposition) and destruction of the portal microvasculature/capillaries (32,48–50) branching off the communicating artery from which periductal vascular plexus arises (49), resulting in ischemic bile duct injury and loss.
Humoral mechanisms of alloreactivity mediated by donor-specific HLA antibodies appear to frequently operate along with cellular mechanisms of rejection. This may explain the general benefit of including mycophenolic acid derivatives in the prevention and treatment of liver rejection, as it inhibits B-cell proliferation and antibody production. Evidence exists that the commonly used CNI tacrolimus and cyclosporine also affect antibody formation through interference with Th cells essential for the B cell-mediated immune responses (1,51). This aspect helps explain the response in most cases of mixed cellular and HR of the liver to current immunosuppressant therapy directed primarily against T cell. (52). Further studies are required to understand the level of immune dysregulation of alloantibody production responsible for the rejection refractoriness to the usual immunosuppressive therapy seen on occasion as progressive cholestasis due to ongoing bile duct damage and loss.
Our study is limited by the biases inherent to any retrospective survey. A selection bias is introduced by the non-consecutive nature of our cases. This bias is reduced by having most degrees of clinical and histological severity of rejection represented in patients with and without DSA, as well as cases without rejection. Due to the absence of protocol biopsies (which are not performed at our institution) and surveillance DSA determination, we are unable to determine the presence of subclinical AMR, as our subjects were selected to undergo liver biopsy indicated by liver dysfunction. Our study does not allow accurate gauging of the possible clinical significance of focal C4d deposition or of circulating DSA without associated C4 deposition. Because of the very limited number of follow-up biopsy with C4d staining, we can not provide data about the duration of C4d persistence, or the dynamics of DSA after rejection treatment and return of liver tests to baseline. As the C4d binds to the nearby proteins or carbohydrates covalently, it is expected that it takes the half-life of those substances, and hence to have a variable tissue persistence. Serial biopsies will be required to determine the dynamics of C4d deposition. Our study population is skewed toward patients with acute cellular and chronic ductopenic rejection. We believe that a properly resourced prospective study will be necessary to accurately assess the spectrum of humoral alloreactivity and to allow standardization of detection methodologies in liver transplant recipients, while limiting the confounding influence of biases that are inherent in a retrospective study.
In summary, it appears that humoral alloreactivity mediated by antibodies against donor HLA molecules is frequently intertwined with cellular mechanisms in the process of rejection of ABO identical/compatible liver allografts. Our study also supports the view that AMR in liver transplantation has a role in the destruction of the portal microvasculature leading to ischemic injury of the interlobular bile ducts, and ductopenia. This is particularly critical when it develops in the early posttransplant period as rapidly progressing cholestasis with a potentially inexorable course may endanger the liver allograft viability unless therapy specifically directed toward antibody depletion is initiated. An important conclusion of our study is that there is a need for prospective studies to correlate the detection of circulating DSA with C4d tissue deposition, histology and outcome of the liver allograft in order to advance our understanding of the significance of AMR after liver transplantation.
The authors would like to thank Mrs. Barbara Voss for her invaluable help with this project.
Disclosure: The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.