Antibody-mediated rejection in liver transplantation: Current controversies and future directions



Interest in the role of donor-specific human leukocyte antibodies in liver transplantation has been rekindled recently. Emerging evidence suggests that these antibodies may cause injury to the liver allograft. Here we review the clinical literature, highlight controversial results, and propose a path forward for the definition and better understanding of antibody-mediated injury to the liver. Liver Transpl 20:514–527, 2014. © 2014 AASLD.

Donor-specific human leukocyte antigen alloantibodies (DSAs) have attracted significant interest in solid organ transplantation in the past few years, and they are widely accepted as a risk factor for decreased graft survival after kidney and heart transplantation. In contrast, the liver appears relatively resistant to DSA-mediated injury. Indeed, most liver transplant programs pay little attention to DSA levels. Furthermore, simultaneous liver transplantation with another organ has been reported to facilitate the absorption of circulating DSAs and protect the second organ from injury.

Recently, more detailed studies have begun to elucidate the impact of DSAs on the liver and question the notion that the liver is always spared from DSA-mediated injury. With the introduction of solid-phase assays, which allow detailed semiquantitative analyses of human leukocyte antigen (HLA) antibodies, several studies have suggested that, in some instances, DSAs can persist after liver transplantation and may be associated with adverse outcomes. The goals of this overview are to assess the current evidence regarding DSAs and liver transplantation and to highlight the areas that require further study.


In 1969, Patel and Terasaki[1] published their landmark article demonstrating the association between a positive cross-match and an increased risk of hyperacute rejection of the kidney allograft, which sets the stage for antibody-mediated graft injury; this was later termed the humoral theory of transplantation.[2, 3] Because of the strong clinical evidence, shortly thereafter, preoperative cross-matching of the recipient's serum and the donor's lymphocytes became a requirement of kidney transplant programs throughout the world. The earliest assay was complement-dependent cytotoxicity (CDC) cross-matching of the recipient's serum and the donor's lymphocytes. Although this assay can effectively predict hyperacute rejection, its sensitivity and specificity (due to non-HLA antibodies) are low. Flow cytometry (FC) cross-matching relies on the detection of antibodies binding to the surface of donor lymphocytes, and it is more sensitive than CDC cross-matching. Unlike these cell-based assays, current solid-phase immunoassays such as Luminex single-antigen beads (SABs) enable the detection of HLA antibodies specifically and semiquantitatively[4, 5] and allow the performance of a virtual cross-match.

In addition to causing hyperacute rejection, HLA antibodies can induce a slower and prolonged pattern of injury to the kidney graft. In fact, several large prospective studies have demonstrated that graft survival is significantly decreased in the presence of circulating DSAs.[2-5] Even low levels of DSAs are a risk factor for the development of transplant glomerulopathy in cross-match–positive kidney recipients.[6] These detrimental effects of HLA antibodies are thought to be produced through 3 mechanisms: (1) complement activation through the classic pathway that leads to the formation of the membrane-attack complex; (2) direct injury to the capillary endothelium through antibody interactions with cell surface antigens; and (3) the activation of proinflammatory cells that express fragment crystallizable receptor, such as natural killer cells, macrophages, and neutrophils.[7]

Because of the overwhelming evidence for antibody-mediated injury to kidneys, a consensus conference was held in 2003 to define the diagnostic criteria for antibody-mediated rejection in solid organ transplantation. This group recommended 4 diagnostic criteria. Accordingly, the diagnosis of antibody-mediated rejection requires evidence of graft dysfunction, tissue injury, antibody action [complement component 4d (C4d) or immunoglobulin deposition], and circulating HLA antibodies at the time of biopsy.[8] Although the acceptance of these criteria standardized the diagnosis of antibody-mediated rejection of kidney and heart allografts, whether such a phenomenon exists in liver transplantation remains unclear.


Histological examinations of some of the earliest liver allografts demonstrated antibody and complement deposition on the endothelial surface, but this deposition appeared to be much less intense and less frequent than that seen in kidney and heart allografts.[9] The Pittsburgh group reported its initial experience with cross-match–positive liver transplantation in 1981.[10] In 11 recipients with a strongly positive lymphocytotoxic cross-match, they observed no hyperacute rejection. This finding was later extended to a larger cohort of 667 patients who underwent liver transplantation between 1980 and 1985. Sixty-seven patients (10%) had a panel-reactive antibody value > 30%, and graft survival and patient survival were not affected by the preoperative panel-reactive antibody status.[11] In those relatively early days of liver transplantation, the standard immunosuppression was cyclosporine A and steroids, and the 5-year overall survival rate was approximately 50%. The cytotoxic cross-matches were done retrospectively; they were investigative in nature and did not influence the clinical decision. As the number of liver transplants increased, the criteria for early rejection [ie, acute cellular rejection (ACR)] and chronic rejection (CR) of the graft were defined. In 1987, immunoglobulin M and immunoglobulin G (IgG) deposition on the small and medium arterioles of chronically rejected liver allografts was first demonstrated,[12] and this raised the possibility of a role for antibodies in chronic injury to the liver.

For the next decade, as the focus was shifted primarily toward improving strategies to control the cellular component of alloimmunity, interest in the role of HLA antibodies in liver transplantation subsided. This trend continued until solid-phase immunoassays detecting circulating HLA antibodies became widely available.


The impact of allosensitization on the incidence of ACR has been addressed by several studies (Table 1). Interestingly, the findings of these studies have been variable. For example, in a study from 2001 involving 91 patients with a positive pretransplant FC cross-match, the incidence of acute rejection in the first year was similar to that for 374 patients with a negative cross-match (57% versus 53%).[15] However, in this single-center experience, patients with higher levels of DSAs, as determined by FC mean fluorescence intensities, had a higher incidence of steroid-resistant rejection (31% versus 4%). In the study, these rejection episodes were all successfully treated with OKT3, and graft survival at 1 year was not inferior to that of cross-match–negative recipients. Similarly, in a retrospective study of 48 liver transplant recipients, the preoperative cross-match status, regardless of the method used (cytotoxicity or FC), did not correlate with the incidence of acute rejection within the first year.[17] Furthermore, the 1-, 3-, and 5-year overall survival rates were comparable for the 2 groups.

Table 1. Literature on HLA Antibodies and Their Impact on Liver Transplantation
 Patients (n)  Detection of HLA Antibodies   
Study (Year)Positive for HLA AntibodiesNegative for HLA AntibodiesTransplant TypeFollow-UpMethodTimeMaintenance ImmunosuppressionProtocol BiopsyFindings
Kasahara et al.[13] (1999)553LDLT3 monthsFC cross-matchBefore transplantationCNI + steroidNoNo difference in allograft outcomes for cross-match–positive recipients
Takaya et al.[14] (1999)28178DDLT7.7 yearsT cell CDC cross-matchBefore transplantationCNI ± steroidNoIncreased incidence of biliary complications and CR in cross-match–positive recipients
Scornik et al.[15] (2001)91374DDLT1 yearFC cross-matchBefore transplantationCNI + steroid + AZTNoNo difference in incidence of ACR; more steroid-resistant ACR for cross-match–positive recipients
Bishara et al.[16] (2002)1844DDLT1 yearCDC-AHG cross-matchBefore transplantationCNI + steroidNoIncreased incidence of ACR and graft loss for cross-match–positive patients in first 3 months
Matinlauri et al.[17] (2005)1632DDLT5 yearsCDC cross-match + FC cross-matchBefore transplantationCNI + steroid + AZTNoNo increased incidence of ACR; no difference in survival at 1, 3, and 5 years
Muro et al.[18] (2005)14254DDLT>5 yearsT cell CDC + HLA class I SABBefore transplantationCNI + steroid + AZTNoNo increased incidence of ACR or CR but markedly decreased graft survival
Castillo-Rama et al.[19] (2008)89755DDLT>5 yearsCDC cross-match (T cell, B cell, and mixed) + SABBefore transplantationCNI + steroid ± MMF or AZTNoShorter graft survival with T cell CDC cross-match positivity or class II HLA antibodies (non-DSAs); no association with CR
Goh et al.[20] (2010)33106DDLT (all retransplants)4 yearsCDC cross-match (mixed lymphocyte) + SABBefore transplantationNot reportedNoShorter graft survival for retransplant recipients with class I HLA antibodies; no difference for those with class II HLA antibodies or any DSAs
Musat et al.[21] (2011)2716DDLT + LDLTNot availableSABAfter transplantationCNI + steroid ± MMFNoIncreased incidence of ACR and ductopenia in recipients with concomitant circulating HLA antibodies
O'Leary et al.[22] (2011)3939DDLT>10 yearsSABBefore and after transplantationVariable, mostly CNI + steroidNoHigher incidence of de novo (but not preformed) DSAs in recipients with CR
Lunz et al.[23] (2012)100709DDLT + LDLT>5 yearsCDC-AHG cross-match ± SABBefore transplantationCNI (± steroid in cross-match–positive patients)YesNo increased incidence of ACR; no difference in graft survival for cross-match–positive recipients
Taner et al.[24] (2012)2070DDLT + LDLT1 yearFC cross-match + SABBefore and after transplantationCNI + steroid + MMFYesNo increased incidence of ACR; no difference in graft survival for cross-match–positive recipients
Miyagawa-Hayashino et al.[25] (2012)3247LDLT>10 yearsSABAfter transplantationCNIYesAssociation of graft fibrosis and concomitant class II DSAs
Ruiz et al.[26] (2012)2732450DDLT≤20 yearsCDC cross-match (T cell and B cell)Before transplantationNot reportedNoNo difference in patient or graft survival for cross-match–positive recipients
Shin et al.[27] (2013)801053DDLT + LDLT≤5 yearsCDC cross-match (T cell and B cell)Before transplantationCNI + steroid + MMFNoNo difference in patient or graft survival but increased incidence of liver-specific graft loss, which includes recurrent viral disease and technical losses due to hepatic artery and portal vein thrombosis
O'Leary et al.[28] (2013)1841086DDLT≥4 yearsSABBefore and after transplantationVariable, mostly CNI + steroidYesIncreased incidence of early ACR for patients with class II DSAs; shorter 5-year graft survival for patients with class I and/or class II DSAs

Another single-center, retrospective analysis of 844 patients (89 of whom had a positive cross-match at the time of transplantation) suggested an increased incidence of acute rejection in recipients with preformed HLA class II antibodies but not in those with class I antibodies.[19] The 5-year graft survival of patients with preformed HLA class II antibodies was also lower (72% versus 56%). Anti-HLA class I antibodies were associated with decreased survival rates within the first year, but this association was lost after the first year. Similarly, another group observed early acute rejection (in the first 3 months after transplantation) more frequently in cross-match–positive recipients.[16] However, the incidence of acute rejection beyond the first 3 months was not different, and graft survival within the first year was not inferior. In contrast, a pretransplant positive FC cross-match did not increase the incidence of acute rejection in the first 3 months in pediatric living donor liver transplantation (LDLT).[13]

In the largest series to date, an association between early cellular rejection (less than 6 months after transplantation) and preexisting HLA class II antibodies was seen in 1270 recipients, of whom 184 (14.4%) had preexisting DSAs.[28] In a prospective analysis of 90 consecutive liver transplants, we found that the incidence of a positive cross-match was 24%.[24] According to protocol biopsy samples obtained from all the patients in this cohort (which allowed us to look for both clinical and subclinical cellular rejection), there was no association between the cross-match status and the incidence of acute rejection within the first year.

Because of these conflicting results, it is difficult to draw conclusions. Many of these studies are from a different immunosuppression era, the overall numbers of cross-match–positive recipients are low, and the follow-up is relatively short (mostly 1-year graft survival). If studies from the recent era of immunosuppression and DSA detection are studied specifically, one might conclude that DSAs, especially against class II, may represent an increased risk of early acute rejection, but short-term graft survival is unaffected. Thus, DSAs do not appear to have the catastrophic consequences of hyperacute rejection leading to graft loss that can be observed in kidney transplantation.


The initial observations regarding the potential role of HLA antibodies in liver CR came from the Pittsburgh group. Allografts with pathological evidence of CR had antibody deposits in the arterioles.[12] Earlier clinical studies suggested that CR was much more frequent in presensitized patients (26.3%) versus nonsensitized patients (4.9%).[14] Vanishing bile duct syndrome, a term often used synonymously with CR in those days,[29] was first thought be the result of circulating anti-HLA antibodies. With further analysis, the development of this histopathological finding was determined to be related to inadequate immunosuppression and incompletely treated cellular rejection,[30, 31] and the term is no longer used widely. In 2005, a retrospective analysis of 268 consecutive liver transplants demonstrated that the incidence of CR was not different in cross-match–positive recipients versus cross-match–negative recipients.[18] Despite this, graft survival was markedly decreased in cross-match–positive recipients with a 1-year survival rate of merely 28.6%, although there were only 11 patients in this group. Most of the graft losses in this cohort were attributed to sepsis.

Recently, the Baylor group published a series of studies of their patients.[22] When 39 liver transplant recipients with CR were compared to the same number of case-matched control patients without CR in a retrospective analysis, DSAs were observed more frequently in the former group. For this analysis, the control group was chosen from recipients with no history of acute rejection, whereas 87% of the members of the CR group had at least 1 episode of biopsy-proven ACR. Furthermore, a significant percentage of the control group had preformed DSAs (41% versus 60% in the CR group). The mean fluorescence intensities of class I DSAs, but not class II DSAs, were higher in the CR group. In a subsequent publication,[32] the same authors reported that multiple IgG subclasses were found in the sera of CR patients, and the IgG3 subclass was associated with an increased risk of graft loss (hazard ratio = 3.35). These analyses were done only for posttransplant DSAs, and the presence of multiple IgG subclasses was 4-fold higher for recipients from whom steroids were withdrawn more than 1 month after transplantation. In a comprehensive review of their entire 20-year experience with 2723 liver transplants from 1987 to 2007,[26] however, the same group recently reported no increased incidence of rejection in cross-match–positive recipients.


Similarly to the conflicting reports about the rates of liver allograft rejection in presensitized patients, studies investigating chronic graft injury and overall outcomes have thus far come up with variable results. The first potential correlation between a positive T cell cross-match and graft injury was reported in 1999.14 In this retrospective analysis, the incidence of biliary complications during the average follow-up of 7 years was found to be 36.8% for cross-match–positive recipients and 10.2% for cross-match–negative recipients. According to a more recent analysis of a single-center experience,[21] circulating DSAs were documented in 7 of 10 liver transplant recipients with biopsy-proven ductopenia. The ductopenia was attributed to inadequate immunosuppression in 7 patients, whereas the other 3 patients had preceding ACR. Whether these patients had preexisting DSAs or developed de novo DSAs (ie, because of inadequate immunosuppression) was not discussed. The development of de novo DSAs (with a mean fluorescence intensity > 5000) in 61 recipients (8.1%) was associated with decreased 5-year graft and patient survival in a recent retrospective analysis of 749 adult liver transplant recipients.[33]

In a more detailed study of 809 liver transplants from a single center,[23] allograft biopsy samples from cross-match–positive recipients had more histological evidence of biliary duct, portal, and venous subendothelial inflammation. Biopsy samples were obtained more than 3 weeks after transplantation from 9 recipients with persistent circulating DSAs, and ongoing ACR or ischemic cholangitis was found in 7 of them. After excluding all confounding factors, the authors estimated that the incidence of isolated antibody-mediated rejection in cross-match–positive patients was approximately 5%. Despite the histopathological findings, however, in this subgroup, the liver enzyme and bilirubin levels were not different from those of cross-match–negative recipients, and the graft and recipient survival rates were similar.

Recent large retrospective analyses from the United States,[26] Asia,[27] and Europe[20] have shown that presensitization has no impact on graft or recipient survival. According to a 20-year follow-up study of 2723 consecutive transplants,[26] the rates of B cell and T cell cross-matches were 10.1% and 7.4%, respectively, and no significant differences in the incidence of vascular or biliary complications were observed. In an Asian series of 1133 liver transplants (mostly LDLT),[27] the incidence of positive cross-matches was 7.1%. Although rejection rates were comparable, more grafts in the cross-match–positive group were lost to a combination of liver-specific reasons, including recurrent viral hepatitis and hepatic artery and portal vein thrombosis.[27]

Among pediatric LDLT recipients, the incidence of inflammation and the severity of fibrosis were found to be higher in those with circulating DSAs.[26] However, these findings were not always predictable because some recipients with high levels of circulating DSAs had normal graft histology.


Several histological changes found in presensitized liver transplant recipients have been described.[34-36] These changes include centrilobular hepatocyte swelling, hepatocanalicular cholestasis, and acute cholangiolitis in the early period (up to 1 week after transplantation). Portal edema and inflammation accompany these changes after a period (1-4 weeks after transplantation) and may eventually lead to thrombosis of large vessels and ischemic cholangiopathy. None of these changes, however, are unique to presensitized patients because similar histopathological findings are observed in ischemia/reperfusion injury, hepatic artery thrombosis, or biliary ductal obstruction. Complicating the picture further is the fact that cellular rejection often accompanies these changes.[35]

In kidney and heart transplantation, C4d is now widely used as supporting evidence for antibody-mediated injury. C4d is a split product of complement activation with no known biological function.[37] Unlike other complement factors that bind to tissues with weak interaction, C4d binds to tissues covalently and, therefore, is easier to recognize histologically. Linear C4d staining in peritubular capillaries was initially found to be related to graft loss in kidney transplantation.[38] This was later confirmed in other kidney recipients with suspected antibody-mediated rejection and circulating DSAs.[39] In 2003, C4d staining was incorporated into the Banff criteria for antibody-mediated kidney rejection. Recent evidence suggests that C4d may be a better indicator of chronic injury than DSAs.[40]

There have been significant efforts to use C4d staining in liver transplantation as a surrogate marker for antibody-mediated injury (Table 2). In comparison with kidney transplantation, this has been more difficult for both technical and liver-specific reasons. C4d immunohistochemistry appears to yield more specific results for fresh frozen biopsy specimens; however, this is often not practical, and as a result, most of the published reports have used paraffin-embedded tissue blocks. In addition, the liver possesses numerous compartments that, under different conditions and disease states, can enhance C4d deposition. For example, the staining of hepatocytes with C4d antibodies is observed as a result of ischemia/reperfusion injury,[41] and it is not an antibody-mediated injury marker. Diffuse portal C4d staining can be seen in up to two-thirds of biopsy samples with ACR,[21, 42] although a recent report demonstrated that there is no difference between cases of ACR with or without diffuse C4d deposition with respect to the Banff grade of rejection or liver enzymes.[23] Several studies have looked at the differences in the location of C4d deposition between DSA-positive and DSA-negative liver transplant recipients. Arterial, portal venous, and sinusoidal endothelial C4d staining is more frequently observed in cross-match–positive recipients.[41, 42] This deposition seems to occur readily because sinusoidal C4d staining can be seen 2 hours after reperfusion in cross-match–positive patients.[43] Within 21 days, diffuse C4d staining is significantly more common in cross-match–positive recipients. In day 7 protocol biopsy samples, we observed that diffuse C4d deposition in all compartments of the liver was seen only in recipients with concomitant circulating DSAs, regardless of the preoperative cross-match status.[24]

Table 2. Literature on the Utility of C4d Immunostaining of Liver Allograft Biopsy Samples From Recipients With Circulating HLA Antibodies
Study (Year)Patients (n)TissueAntibodyFindings for Recipients With HLA AntibodiesFindings for Recipients With No HLA Antibodies
Bellamy et al.[41] (2007)101Paraffin-embeddedMonoclonalDiffuse portal venular and sinusoidal stainingPortal venular or sinusoidal staining
Sakashita et al.[42] (2007)198Paraffin-embeddedPolyclonalDiffuse portal venular, portal stromal, and lobular stainingPortal staining in cases of CR, recurrent viral hepatitis, and primary biliary cirrhosis
Musat et al.[21] (2011)43Paraffin-embeddedPolyclonalDiffuse portal capillary and portal stromal stainingDiffuse portal staining in some with biliary strictures
Lunz et al.[23] (2012)297Paraffin-embeddedPolyclonalDiffuse portal stromal, portal endothelial (hepatic artery, portal vein, and capillary), sinusoidal, and central venous stainingDiffuse/focal portal staining in some with biliary strictures and recurrent viral hepatitis
Kozlowski et al.[43] (2012)68Paraffin-embedded +fresh frozenMonoclonalLinear portal stromal, portal endothelial, sinusoidal, and central venous stainingNo staining by immunofluorescence on fresh frozen samples
Ali et al.[44] (2012)53Paraffin-embeddedPolyclonalLinear portal capillary and sinusoidal stainingLinear portal capillary and sinusoidal staining in cases of ACR, biliary strictures, and recurrent liver disease
Miyagawa-Hayashino et al.[25] (2012)79Paraffin-embeddedPolyclonalDiffuse endothelial (hepatic artery, portal vein, and portal capillary) stainingNo diffuse staining
Taner et al.[24] (2012)90Paraffin embeddedPolyclonalDiffuse portal venular, portal stromal, and central venous stainingNo diffuse staining

In LDLT, diffuse perivenular, sinusoidal, lobular, and stromal C4d staining was more often observed in cross-match–positive recipients,[42] although this does not seem to correlate with the incidence of acute rejection or CR.[25] Among pediatric LDLT recipients with protocol biopsy samples obtained at least 10 years after the operation, no significant difference in C4d deposition was observed between the recipients with circulating DSAs and the recipients without circulating DSAs.

Thus, although the published literature is confusing, it appears that linear C4d staining of the microvessels and stroma in more than 50% of the portal tracts and/or sinusoids may be useful for the diagnosis of antibody-mediated injury to the liver if it occurs in the presence of circulating DSAs (Fig. 1). In the absence of DSAs, however, C4d deposition is likely not specific to HLA antibodies. In fact, in the absence of circulating HLA antibodies, C4d staining has been reported in liver allografts with CR (50%), recurrent HCV (50%), autoimmune hepatitis (75%), biliary obstructions or strictures (37%), preservation injury (18%), vascular thrombosis (43%), and ischemic injury (54%) and even in biopsy samples with no diagnostic abnormalities (40%).[44] It is also possible that C4d activation may be triggered by non-HLA DSAs.[21, 45]

Figure 1.

C4d immunohistochemistry: (A) diffuse sinusoidal and (B) portal deposition of C4d as seen in patients with high levels of circulating DSAs.


The deleterious effects of antibodies on blood-type antigens (isoagglutinins) were recognized early on and have been well documented since.[46] These antibodies cause a severe form of rejection, characterized by extensive endothelial damage and hemorrhagic necrosis, in ABO-incompatible liver allografts. The recent success with these transplants due to aggressive perioperative isoagglutinin-targeting therapy with plasma exchange, splenectomy, and B cell depletion have left no doubt about the injury incurred by the liver.[47, 48]

In contrast, conventionally, the liver allograft has been thought to be relatively spared from the effects of preexisting HLA antibodies. Over the years, several mechanisms for this immune privilege have been postulated (Fig. 2). Among the other commonly transplanted solid organs, the liver is unique with its vast vascular (sinusoidal) endothelial surface, which enables the absorption of circulating antibodies. For example, in the rat model, DSAs are cleared from the circulation in only 30 minutes when serum is perfused through an extracorporeal liver of donor origin.[49] Thus, the commonly observed resolution of circulating DSAs after a liver transplant may be simply a spatial phenomenon. Accordingly, the risk of antibody-mediated injury increases as the size of the liver allograft is reduced in rats.[50] The liver also secretes soluble HLA I that can bind to and inactivate corresponding antibodies[51] with the aid of its resident phagocytic Kupffer cells. Furthermore, the liver possesses an unmatched capacity to regenerate, even after a substantial immune-mediated hepatocellular injury, and this is perhaps more important than any other mechanism. These mechanisms, in combination or separately, probably account for the findings in the aforementioned large series with consecutive liver transplants demonstrating no statistical differences in graft outcomes between cross-match–positive patients and cross-match–negative patients. In our experience, the preexisting DSAs were cleared from the circulation within a week of the transplant in 90% of presensitized recipients.[24] This mopping effect of the liver appears to be more effective against class I HLA antibodies,[52] possibly because of the higher levels of class I expression on the liver's sinusoidal endothelial cells. Auxiliary partial liver grafts from the same donor were used to facilitate living donor kidney transplantation in 7 highly sensitized patients, 5 of whom were reported to have cleared the circulating DSAs and have good kidney allograft function after 2 years.[53] Why a minority of cross-match–positive patients do not clear preexisting circulating antibodies, whether this can be predicted at the time of transplantation, and whether these antibodies will lead to graft dysfunction in time remain unknown.

Figure 2.

Proposed mechanisms of liver allograft resistance to DSA-mediated injury.

In contrast, the development of de novo HLA antibodies after liver transplantation may be easier to predict. Recent evidence shows that the formation of HLA antibodies happens as a result of ongoing inflammation in the graft. For example, proinflammatory events such as infections and surgery increase the mean fluorescence intensity and breadth of DSAs and the strength of the cross-match,[54] presumably because of the antigen-nonspecific stimulation of memory B cells or plasma cells by proinflammatory cytokines. This appears to be the case in the development of de novo DSAs in kidney transplant recipients. In nonsensitized kidney transplant recipients who were maintained on calcineurin inhibitor (CNI) and mycophenolic acid–based immunosuppression, the incidence of de novo DSA formation in the first 6 months (the period with the highest adherence to the immunosuppressive regimen) was found to be zero.[55] Thereafter, the prevalence of de novo DSAs increased to 15% to 19% within 5 years,[55, 56] with nonadherence and preceding clinical rejection being the most predictive events for de novo DSA formation.[55] Thus, it is possible that prolonged or repeated alloantigen exposure secondary to tissue injury due to rejection and/or inflammation acts as an immune booster and promotes DSA production after liver transplantation. Among kidney transplant recipients with biopsy-proven antibody-mediated rejection, those who responded to treatment with T cell depletion with or without plasmapheresis had significantly better long-term graft survival than nonresponders.[57] Interestingly, two-thirds of the responders did not receive any therapy aimed directly at reducing circulating antibodies, such as plasmapheresis, and this highlights the role of cellular immunity underlying the clinical diagnosis of antibody-mediated rejection. Similarly, non-HLA IgG autoantibodies, which are detected in liver transplant recipients at a much higher rate in comparison with the general population, respond to steroids or T cell–targeting therapy.[58] The results of a recent analysis confirm these observations in liver transplant recipients because low levels of CNIs and the use of cyclosporine (versus tacrolimus) were found to be associated strongly with the development of de novo DSAs with a mean fluorescence intensity > 5000.33

Because of these spatially and temporally close interactions between the cellular and humoral mechanisms of alloimmunity, it has been difficult to prove that HLA antibodies alone can lead to liver allograft injury. Multiple retrospective analyses, as discussed previously, have demonstrated an association between the presence of circulating DSAs and the development of CR. This is not an unexpected observation because the most common risk factor for both events is preceding episodes of multiple acute rejections or inadequately treated acute rejection.[59] Without a prospective and detailed analysis of the timeline of events leading to CR, the current evidence does not provide evidence for a causative role for HLA antibodies in graft injury. Spatially, however, the argument appears stronger. Although eventually the hallmark bile duct atrophy develops, fibrosis in CR originates from the perivenular regions to which HLA antibodies have immediate access, away from the portal inflammation of cellular rejection. In fact, the transition from acute rejection to CR in liver allografts has been reported to appear after a decrease or resolution of portal inflammation but with concomitant perivenular inflammation.[59] Thus, it is possible that the inflammation leading to CR is triggered by circulating DSAs (eg, by the activation of effector cells through fragment crystallizable receptor ligation). Genetic and phenotypic markers of one such cell type (natural killer cells) were recently found in kidney recipients with biopsy-proven antibody-mediated rejection and concomitant circulating DSAs, and this provides evidence for the role of these cells in the injury prompted by the antibodies.[60] Interestingly, centrilobular (perivenular) fibrosis in late biopsies (>5 years after transplantation) is not uncommon for pediatric LDLT recipients with circulating DSAs, and it may be reversible through an increase in immunosuppression.[25, 61, 62]

If this mechanism of action is indeed effective, then extra caution must be observed for recipients who are enrolled in immunosuppression minimization or withdrawal trials. In a retrospective review of patients who had been off immunosuppression, those with circulating DSAs were found to be more likely to develop rejection.[63] Similarly, the results of a recent prospective immunosuppression withdrawal trial of pediatric liver transplant recipients demonstrated that recipients with portal inflammation and increased C4d deposition at the time of trial entry were more likely not to become tolerant, although preexisting DSAs at the time of withdrawal did not influence the outcome.[64] Importantly, a number of recipients in both the tolerant group and the nontolerant group developed de novo DSAs after weaning from immunosuppression. Whether these observations may be generalized to adult liver recipients remains to be seen.


The current literature suggests that DSAs can have deleterious effects on liver allografts; however, their impact appears to be less severe than that observed in kidney transplantation. In order to better understand the impact of DSAs on both early and late outcomes of liver transplantation and to reconcile different clinical experiences, we suggest that longitudinal and prospective studies, akin to those in kidney transplantation, correlating DSAs (both before and after transplantation at routine intervals) with clinical outcomes and graft histology through protocol biopsies are needed. These studies would ideally be designed to analyze specific factors in detail, such as antibody levels, the ability of antibodies to bind complement, antibody specificity (anti-HLA class I versus class II), target antigen expression in the allografts, immunosuppressive regimens, and compliance issues in the recipients. All of these factors appear to be important in kidney transplant recipients; however, their role in liver transplantation is still somewhat unclear. Similarly, in kidney allografts, microvascular inflammation (infiltration of the peritubular capillaries and glomeruli with natural killer cells, macrophages, lymphocytes, and dendritic cells) has been shown to correlate with chronic antibody-mediated injury.[65] In addition, early signs of endothelial cell activation can be detected by electron microscopy in otherwise normal-appearing renal allografts that later develop antibody-mediated injury. No such detailed studies exist for liver transplant recipients with circulating DSAs. As discussed previously, a gene expression profile that appears relatively specific for antibody-mediated kidney rejection has been identified and has been suggested to be more valuable than graft histology in diagnosis; it is possible that a similar signature may be identifiable for liver transplants suffering antibody-mediated injury. Finally, in sensitized recipients, both the cellular immune response and the humoral immune response may become activated. Thus, a study of the role of the sensitized cellular response may lead to important insights into the outcomes of liver transplant recipients (Table 3).

Table 3. Currently Available Data for DSA-Associated Injuries in Kidney Transplantation and Liver Transplantation
 Kidney TransplantationLiver Transplantation
DSA specificity and levelsAbundant data[1, 40, 66]Increasing data
C4d deposition in microvasculatureAbundant data[37, 38]Increasing data
DSA subtypes, C1q bindingEmerging data[67-69]Limited data
Microvascular inflammationEmerging data[70-72] (eg, peritubular capillaritis)No analogous data
Endothelial cell activation by light microscopyAbundant data[73-75] (eg, transplant glomerulopathy)No analogous data
Endothelial cell activation by electron microscopyLimited data[76]No analogous data
Gene expression profile of chronic antibody-mediated rejectionEmerging data[77-79]No analogous data
Therapeutic trials designed to prevent DSA-associated injurySeveral studies[80-82]None


Recent evidence suggests that DSAs may play a role in graft injury and overall outcomes of liver transplantation, although this impact appears to be less common and less severe than that seen in kidney and heart transplantation. A better understanding of this phenomenon through prospective studies is warranted to prevent graft injury in this subset of highly sensitized liver transplant recipients.


acute cellular rejection


anti-human globulin




complement component 1q


complement component 4d


complement-dependent cytotoxicity


calcineurin inhibitor


chronic rejection


deceased donor liver transplantation


donor-specific human leukocyte antigen alloantibody


flow cytometry


human leukocyte antigen


immunoglobulin G


living donor liver transplantation


mycophenolate mofetil


single-antigen bead