Post-transplant lymphoproliferative disorder (PTLD) is a spectrum of major, life-threatening lymphoproliferative diseases occurring in the post-transplant setting. The majority of PTLD is of B-cell origin and is associated with several risk factors, the most significant being Epstein-Barr virus (EBV) infection. EBV's in vitro transforming abilities, distinctive latency, clonality within the malignant cells and response to targeted therapies implicate a critical role in the biology of PTLD. This minireview focuses on EBV-related PTLD pathogenesis, in particular the interplay between aspects of the EBV life cycle and latency with nonviral factors resulting in the wide spectrum of histology and clinical presentations encountered in PTLD. With the increased prevalence of transplantation a rise in the incidence of PTLD may be expected. Therefore the importance of laboratory and animal models in the understanding of PTLD and the development of novel therapeutic approaches is discussed.
Post-transplant lymphoproliferative disorders (PTLD) are a heterogeneous group of lymphoproliferative diseases that arise after solid organ and other transplantations. They represent a spectrum of clinical and histopathological manifestations ranging from a benign self-limited form of lymphoproliferation to an aggressive, widely disseminated lymphoma (1). The incidence of PTLD relates to the type of transplanted organ (which also influences the amount of lymphoid tissue transferred), intensity of immunosuppression and underlying disease. These risk factors are reflected in the highest incidence of PTLD occurring after small bowel and multiple organ transplants (of the order of 5–20%), followed by thoracic allografts (2–10%) and then 1–5% for renal and liver transplants (2). Although the occurrence of PTLD decreases after the first post-transplant year, the cumulative incidence increases with the time elapsed since transplantation. Therefore the younger the allograft recipient, the greater the risk of developing PTLD over the course of their lifetime.
Approximately 50–80% of PTLD biopsies are positive for Epstein–Barr virus (EBV) within the tumor cells (EBV-related PTLD). Reflecting the critical role of reduced cellular immuno-surveillance against EBV in its pathogenesis, other risk factors for PTLD are the EBV-serostatus of the donor (D) and recipient (R) (D+/R− are at greatest risk); cytomegaloviral disease and the use of T-cell depleting agents. The incidence of PTLD is highest in the pediatric age group, where it is the most common cancer to develop in children after transplantation (3).Children are frequently EBV-seronegative (in developed countries, 50% of children by the age of 5 years are seropositive). The majority of PTLD seen in the pediatric setting occurs within 6 months of primary EBV infection. In 90% of cases following transplantation, EBV-positive PTLD arises from host lymphocytes (the converse is true following myeloablative allogeneic stem cell transplantation). In EBV-seronegative recipients, recipient cells are infected either by the virus present in the donor organ or by primary infection from a third party; whereas in seropositive recipients the PTLD tissue frequently but not invariably contains the recipient EBV isolate (4).
This review is limited to a discussion of PTLD in the context of EBV. For a comprehensive review of non-EBV-specific therapies (such as reduction in immunosuppression, chemotherapy and monoclonal antibodies), the reader is directed elsewhere (5,6).
As a consequence of its histological diversity, a number of alternative histological classifications have been proposed (7). The most widely used is that of the World Health Organization (8). This integrates PTLD into one category with four subdivisions. Early lesions (reactive hyperplasia and infectious mononucleosis [IM]-like) refer to polyclonal B-cell lymphoid proliferations with an admixture of T cells and preservation of normal tissue architecture. Reactive hyperplasia is characterized by numerous plasma cells and rare immunoblasts, whereas in IM-like PTLD there is marked paracortical expansion and immunoblasts are more prominent. Both are generally observed in EBV-seronegative recipients following primary infection, and typically occur within the first year following transplant.
Polymorphic PTLD represents a unique B-cell histology that is very rarely observed outside the context of other immunosuppression-related lymphomas (1). Morphology is variable, but includes architectural effacement with B-lymphocytes at varying stages of differentiation including immunoblasts, small/intermediate-sized lymphoid cells and plasma cells. Mitotic figures are frequent. Polymorphic clonal B cells are contained within a background of nonclonal B and T cells. Reed–Sternberg cells (more typically observed in Hodgkin Lymphoma) may be seen. In contrast to adults, polymorphic PTLD are the most commonly observed histological subtype in children. Polymorphic PTLD are generally EBV-related. Monomorphic PTLD are individual B- and T-cell lymphomas, which are then subclassified according to the WHO lymphoma classification in the immunocompetent host. The immunophenotype is dependent on the histological subtype, but T and NK lymphomas comprise only a minority of cases (they occur late and are generally EBV unrelated). They are most frequently observed in adults, and the commonest monomorphic entity is diffuse large B-cell lymphoma in which there is replacement of follicularity with diffusely dispersed monoclonal B cells. The next most common monomorphic histology is Burkitt or Burkitt-like. Histological subdiagnosis in PTLD is imperfect, and it is not unusual for areas of polymorphic and monomorphic histology to coexist within the same biopsy, or for discordant findings between biopsies taken from different sites.
The final division is miscellaneous lesions such as Hodgkin-like lymphoma. Indolent B-cell lymphoma histologies (such as follicular lymphoma) are rarely seen in the context of PTLD.
The clinical presentation is variable, depending on the underlying pathologic condition, the type of transplant and the time since transplant (9). Presentation ranges from nonspecific symptoms of weight loss, malaise and fever to more severe organ dysfunction or infectious complications. Symptoms and signs are similar to those seen during primary EBV infection and include fever, sweats, generalized malaise, enlarged tonsils and cervical lymphadenopathy. Median onset of PTLD is approximately 6 months after transplant, but it has been reported from 1 week to 9 years after transplant. Studies have suggested that the monomorphic subtypes of PTLD may have a worse prognosis than polymorphic subtypes; however this has not been observed in more recent studies (10). Early onset PTLD occurring within 1–2 years of transplantation is more common among pediatric recipients and usually consists of an IM-like syndrome with cervical and/or tonsillar enlargement or pyrexia of unknown origin.
The greatest burden to transplant practice is from the polymorphic and monomorphic (i.e. lymphomatous) subtypes (1). In part this is due to reduced chemotherapy tolerance. Relapse is not uncommon, and 90% occur within the first year of remission induction (11). As with EBV-related lymphomas in the immunocompetent, they are commonly extranodal, and the transplanted organ may be involved. Outside the allograft, typical sites of involvement include the liver, gastrointestinal tract, skin and central nervous system. If at all possible, an excision biopsy should be performed. Fine needle aspirates are seldom sufficient and are more likely to lead to diagnostic delay. Assessment and staging are as for lymphomas in the immunocompetent (12). Use of positron emission tomography may allow appreciation of additional extranodal sites not picked up on conventional computerized tomography (13). Although a variety of clinical prognostic scores have been advocated, it remains unclear if any of these are superior to the international prognostic index that was originally devised for lymphomas in the immunocompetent (14).
EBV Viral Load Monitoring
Testing for serological conversion (in the EBV-serongative recipient) is unreliable in patients on immunosuppressive therapy. Nor is it of value in the EBV-seropositive recipient. By contrast, the elevated EBV viral load (as quantified by PCR of EBV-DNA) in the peripheral blood of transplant patients is an established risk factor for EBV-related PTLD (9,15). Serial viral load monitoring of transplant patients is practised routinely in some centers. This enables early or preventative treatment and appears to have decreased the frequency of EBV-related PTLD patients at participating centers (16). The role of viral load monitoring as an indicator of disease response in established PTLD is less clear. For example in patients treated with rituximab the disappearance of cellular viral load does not predict clinical response (17). In addition, recurrence of EBV-related PTLD is not always preceded by a rise in viral load. One criticism of viral load monitoring is the lack of consistency between methods and also the nature of the peripheral blood specimen used (whole blood, cell-free or cellular components), as well as the lack of a universal reference standard and reporting units (18). Simultaneous screening for multiple EBV genes may increase detection rates (15). Although the lack of common methodology makes comparisons difficult, several multicenter, multimethod comparison studies have been conducted and have found that interlaboratory qualitative measurements are comparable (19). Intralaboratory quantitative measurement variation was generally low indicating good assay performance while interlaboratory quantitative measurement variation was reduced with the use of a common collaborator or common extraction method (19,20). The International Working Group for the Standardization of Genome Amplification Techniques is currently working on establishing a universal reference standard, which should greatly aid consistency between centers (21). Ideally, decreasing variability across centers will increase our understanding of the relationship between viral load and clinical manifestations and further assist clinician's interpretation of viral load results.
EBV Life-Cycle and Latency
EBV is a human herpes virus with a 172 kilobase episomal double-stranded DNA genome (22). It benignly infects over 90% of the population, making it one of the most successful of all human viruses. Infection is usually asymptomatic in childhood, but infection in adolescence frequently results in IM. Oropharyngeal infection results in a localized lytic (replicative) infection followed by infection of circulating B cells and amplification of infection as the virus drives the proliferation of latently infected B cells into the systemic circulation. One model holds that infection occurs in naïve B cells within Waldeyer's ring, resulting in short-lived proliferating blasts. The virus subsequently provides signals to drive these blasts along the lines of normal B-cell differentiation into long-lived memory B cells, thereby enabling persistent infection in a nonreplicative (latent) form (23,24). Latent infection is characterized by maintenance of the genome and expression of a limited number of genes (25). The function of specific latent genes is summarized in Table 1 and is discussed in detail elsewhere (25,26). Latent infection of circulating B cells in IM is characterized by the type III latency pattern (or ‘growth phase’),involving expression of six EBV nuclear antigens (EBNA 1, 2, LP, 3A, 3B, 3C) and two integral latent membrane proteins (LMP1, LMP2). EBV-encoded small RNA (EBER)1 and EBER2 and a highly spliced mRNA transcript (BARF) are also expressed but do not code for proteins and their functions remain to be determined. Type III latency B cells are cleared by a potent anti-EBV-specific T-cell immune response (27). From this stage, latently infected B lymphoblasts mimic the steps of naive to memory B-cell differentiation normally associated with a germinal center (GC) reaction. Coincident with the downregulation of EBNA2 (which transactivates the promoters necessary for the other EBNA genes) the virus switches to the type II latency (‘default program’) of EBNA1, LMP1, LMP2 and EBERs. Both LMP1 and LMP2 are constitutively active and independent of ligands provide the necessary T-cell and B-cell signals respectively required for B-cell differentiation. Long-lived EBV infected resting B cells in the circulation have a type 0 latency in which only EBERs but no latent genes (except in some cases LMP2) are expressed. During cell replication, the virus switches to a type 1 pattern (EBERs and EBNA1). The precise mechanisms governing the switch between latent programs remain unknown. Unlike IM, in which symptoms are principally the result of cytokine production from activated T cells (atypical lymphocytes) in the peripheral blood, the lack of latent gene expression in latency 0 allows host cells to escape immunosurveillance by EBV-specific T cells, explaining the asymptomatic nature of persistent EBV infection, reviewed in (22). The life cycle is completed when latently infected B cells in the oropharynx undergo viral replication. This results in viral shedding into the saliva, potentially permitting transmission to another host. The switch from latency to the lytic cycle is incompletely understood, but is in part a function of B-cell differentiation and is more likely to arise in plasma cells. Although expression of the immediate early gene BZLF1 and its product ZEBRA is necessary for the lytic cycle, its expression alone is insufficient to conclusively demonstrate viral replication.
Table 1. Function of the EBV-latent genes expressed in EBV-related PTLD
EBV latent gene
A viral oncogene illustrated by its ability to transform rodent fibroblasts and the inability of recombinant EBV LMP1-deletion strains to transform B cells; responsible for clumping of LCL and expression of markers of B-cell activation such as CD23, CD30 and cell adhesion molecules; induces upregulation of genes involved in protection from the intrinsic (BCL2, A20, BFL1, cIAP2) and extrinsic (cFLIP) apoptotic pathways; induces cellular cytokines IL10, IL6 and IL8; mimics constitutively active CD40 coreceptor; binds members of the TRAF family through its intracellular domain to activate the NFκB, JNK/AP1, JAK/STAT and p38/MAPK signaling pathways necessary for cell transformation; induces hsa-miR-146a which regulates the interferon response pathway, and hsa-miR-155 which regulates NFκB and stabilizes the EBV copy number
Transcribed from two different promoters to either LMP2A and LMP2B, which are homologous apart from the first exon; neither is essential for B-cell transformation; has a key role in abrogating normal B-cell development by suppressing B-cell receptor-mediated proliferation signals by blocking calcium mobilization and tyrosine phosphorylation (and the activation of the lytic cycle); provides the tonic signals required for B-cell survival by co-opting SYK and SRC-family kinases; believed to prolong the half-life of LMP1
DNA binding nucleophosphoprotein required for the replication and maintenance of the viral episome. Antiapoptotic role by binding to USP7 to prevent stabilization of the tumor suppressor gene p53
Essential for B-cell transformation (the P3HR-1 EBNA2 deletion mutant is unable to transform B cells). Transcriptional activator of LMP1, LMP2 and switches EBV promoter usage from Wp to Cp. Functional homolog of intracellular Notch by activating CBF1-bound genes. Transactivates the c-mycprotoncogene.
EBNA-LP is encoded by the leader of each of the EBNA mRNAs; enhances EBNA2-mediated transcription
These genes appear to have a common origin and are tightly associated with the cellular DNA binding protein CBF1 that targets EBNA2 to promoters. This interaction is essential for continued proliferation of LCLs. EBNA3A combines with EBNA3C to repress the pro-apoptotic protein BIM and the tumor suppressor p16INK4A
EBV is divided into two strains (A and B), and then subdivided on the basis of minor genetic differences. It has been proposed that there is differential oncogenic ability between strains. However, our data indicate that viral strains in EBV-related PTLD tissue biopsies are simply reflective of those present in the underlying population (28).
Laboratory and Animal Models of EBV-Related PTLD
In vitro EBV infects nondividing mature B cells with transformation and subsequent proliferation as immortalized lymphoblastoid cell lines (LCL) (22). EBV-transformed LCL are initially polyclonal, but over time become dominated by a single clone. They display a type III latency, similar to that classically displayed in EBV-related PTLD. However, this is probably an oversimplification: more restricted forms of latency are not atypical in late-onset EBV-related PTLD (in which host genetic perturbations may have rendered certain latent genes redundant). Furthermore, different EBER-positive malignant cells within the same PTLD biopsy can display a variety of latency types (29). Both in vitro transformed LCL and ‘spontaneous LCL’ (EBV-infected B-cells outgrown from the tissues of PTLD patients) are frequently used in the laboratory as surrogate models of PTLD, and to stimulate EBV-specific T-cell responses.
A key requirement is for an animal model of PTLD. Following renal transplantation and antirejection immunosuppression, the Cynomolgus monkey develops a lymphocryptovirus (LCV)-related PTLD (30). LCV is a herpes virus of the same subgroup as EBV. EBV is a human pathogen and infection of nonhumans has been challenging. Importantly, infection of the cotton-top tamarin with EBV itself reproduces key aspects of human infection, and has proved useful for studying lymphomagenesis (31). For logistic reasons, mice are the preferred in vivo model for research into PTLD biology. A variety of immunodeficient mouse models of EBV-related PTLD have been developed including coinoculation with the virus and human mononuclear cells, or injection of human LCL or PTLD biopsy tissue (32). More recently, NOD-scid γc−/− or BALB/c Rag2−/−γc−/− mice have been transplanted with human hemopoietic progenitors to reconstitute the human immune system. In these models EBV infection can cause lymphoproliferative disorders in a dose-dependent manner that resemble the lymphomas normally found in immunocompromized patients (33).
Role of Nonviral Factors in PTLD Pathogenesis
The presence of clonal strains of EBV within monoclonal tumors places EBV infection as a likely causal event in EBV-related PTLD. However, the incidence of non-EBV-positive PTLD remains far higher than the incidence of lymphoma in nonimmunocompromized subjects. This emphasizes the importance of nonviral etiological factors, the study of which has been relatively neglected. Nonviral factors are not necessarily only restricted to EBV-negative PTLD. One theory is that the transplanted organ (a common site of lymphoma involvement) is subject to frequent subclinical rejection that might induce a state of chronic antigen stimulation that contributes to a tumorigenic environment (34). The predilection for other extranodal sites suggests a requirement for specific external stimuli, such as cytokines, and polymorphisms in anti-inflammatory cytokines are associated with susceptibility to PTLD (35). A total of 5′ noncoding region mutations of the BCL6 gene are frequent and are associated with monomorphic histology and an aggressive clinical outcome. Lesions without BCL6 mutations are more susceptible to regression with reduction in immunosuppression. PTLD have genomic aberrations common to lymphomas in the immunocompetent such as gains of 8q24, 3q27, 18q21 and loss of 17p13 (36). However deletion at 13q14.3 (encoding for has-miR-15 and hsa-miR-16) occurs with reduced frequency in DLBCL-PTLD than DLBCL in the immunocompetent, suggesting different pathogenetic mechanisms (37). EBV-related DLBCL-PTLD has fewer recurrent lesions than EBV-unrelated DLBCL-PTLD, including gain of 7p, del(4q25-q35) gains of 7q, 11q24-q25. Thus the wide spectrum of lymphoproliferative disorders encountered in EBV-related and unrelated PTLD suggests an interplay in which genetic aberrations interact with viral oncogenes, impaired immunity and chronic antigen stimulation.
Immunosurveillance by host effector adaptive and innate immunity (e.g. cytotoxic T lymphocytes [CTL] and natural killer [NK] cells respectively) plays a critical role in the detection and killing of lymphoma, whereas the ability to evade recognition by CTL is thought to promote cancer survival. In EBV-related PTLD, a significant component of evasion is clearly the result of iatrogenic immunosuppression leading to an absence of EBV-specific cellular surveillance. This results in proliferation of EBV-transformed B cells. However, the current paradigm that it is only immunosuppression that prevents host effector pathways from eliminating tumor cells should be revised to include the active role of EBV-specific mechanisms. Immunoediting involves a dynamic interplay between host immunity and cells with tumorigenic potential and their microenvironment (38). The process is characterized by three distinct phases that represent a continuum in the immunopathogenesis of cancer, termed elimination, equilibrium and escape (Figure 1) (38). This is likely to be relevant to EBV-related PTLD since EBV has been shown to exploit many immune evasive strategies to successfully establish a latent infection in B cells (39). For example LMP1 can block apoptotic signals delivered through the Fas/Fas ligand and TRAIL/death receptor pathways, and also actively impair cis-presentation of its own epitopes (40,41). Additionally, the frequent heavy intratumoral infiltration with CD4+ T cells may be an indicator for the importance of immuno-regulatory factors (42)
There are a small number of reports of PTLD regression following therapy with the nucleoside analogs acyclovir and ganciclovir (GCV); however data are conflicting (43). In most cases antiviral therapy for treating PTLD is unlikely to be of benefit since nucleoside analogs require viral kinases to be converted to active cytotoxic forms. These are only expressed during the lytic cycle. One strategy is to switch the latent infection to the lytic cycle so as to induce tumor cell death (44). Cell lysis may be enhanced by the addition of GCV. Agents that induce lytic cycle in combination with GCV may have therapeutic potential. A phase I/II study reported impressive responses with reasonable tolerability with the combination of arginine butyrate and GCV in refractory EBV-related lymphomas including six PTLD (45).
EBV-seropositive allogeneic donor lymphocyte infusions have been used successfully to treat EBV-related PTLD but the risk of graft versus host disease (GvHD) has prevented widespread uptake (Table 2). One alternative is the use of haploidentical maternal donor lymphocytes which make use of feto-maternal tolerance to reduce the potential for GvHD (46). Another strategy to reconstitute EBV-specific T-cell immunity is via transfer of in vitro generated autologous and allogeneic EBV-specific CTL. These have successfully been used in some cases for the prophylaxis and treatment of EBV-related PTLD (47). Transferred CTL are capable of homing to sites of disease, including the central nervous system, and EBV-specific T-cell immunity is not impaired by concomitant rituximab (48,49). The immunodominant EBV-latent proteins EBNA3A/3B/3C are good targets for adoptive immunotherapy. By contrast LMP1 is subdominant and is known to display a number of polymorphisms within epitope coding regions that influence EBV-specific T-cell generation (28). Contrary to the previous dogma, EBNA1 is not immunologically silent (28). Application of new approaches to generate T cells that target multiple antigens is resistant to immunosuppression, and that harness cognate T-cell help may improve outcomes (28,50). Although attractive, until now the need for an immediate ‘off the shelf’ therapy has limited adoptive immunotherapy's widespread usage. There are a number of alternative approaches to overcome this. One is to transfer EBV-specific chimeric T cells into patients (51). These comprise antibody fragments specific for EBV-latent antigens linked to genes encoding signaling domains of the T-cell receptor (TCR). Along similar lines, given that TCRs are among the most specific biological structures found in nature, they might also serve as excellent candidates for the molecular targeting of antigen. Conserved TCRs are frequently observed in humans, and there is potential to amass banks of ‘off-the-shelf’ EBNA1-specific receptors. Additionally, such TCRs could be artificially enhanced through mutagenesis, thereby creating an even better 3-dimensional fit for their cognate targets. Indeed, preliminary studies using both ‘natural’ and ‘enhanced’ TCRs have shown promise (52).
Table 2. Virus-specific therapy for EBV-related PTLD
Induction of the viral lytic cycle
In EBV-related PTLD predominantly only latent genes are expressed. These comprise only a small portion of the EBV-genome. Replication of latent EBV utilizes the host cells DNA polymerase. Latent EBV infection does not kill the host cell. The switch to the EBV-lytic (replicative cycle) is mediated by the EBV genes BZLF1 and BRLF1 and viral DNA polymerase. This induces the expression of multiple EBV-lytic genes and viral particle formation. Release of viral particles can result in host cell death. Multiple agents exhibit in vitro and/or in vivo EBV-lytic induction ability, including anthracyclines, gemcitabine, histone deacetylase inhibitors and demethylating agents. A clinical trial of sodium butyrate in combination with ganciclovir for EBV-related lymphomas (including PTLD) gave promising results. However, the methylation status of EBV-related PTLD appears variable, potentially limiting the applicability of agents such as 5-aza-cytidine
Antiviral drugs (e.g. ganciclovir)
Ganciclovir is a pro-drug that requires viral thymidine kinase to be phosphorylated to its active form. This enzyme is not expressed by the majority of EBV-infected cells in EBV-related PTLD. Although small cases series have demonstrated efficacy in some cases, there is insufficient evidence to justify ganciclovir in the routine management of EBV-related PTLD
Autologous EBV-specific T cells
Impairment of EBV-specific T-cell immunity is involved in the pathogenesis of EBV-related PTLD. Tumors retain antigen processing and presentation ability, and express multiple EBV-latent antigens including the immunodominant EBNA3A/3B/3C. Reinfusion of in vitro expanded EBV-specific T cells can result in reconstitution of EBV-specific T-cell immunity and induce remission in EBV-related PTLD. Infusions appear to be associated with relatively minor side-effects. To date, T-cell generation protocols have been time consuming and technically challenging, limiting this therapy to specialized centers
Partially matched unrelated allogeneic EBV-specific T cells
Alloreactivity of EBV-specific T-cells generated from healthy EBV-seropositive subjects appears to diminish with prolonged culture. Reinfusion of partially HLA matched unrelated allogeneic EBV-specific T cells into EBV-related PTLD patients does not appear to induce graft versus host disease (GvHD), and induces clinical responses in some patients. Allogeneic EBV-specific T cells can be generated in bulk and cryopreserved and banked in centralized laboratories, to be shipped when needed, providing the option of an ‘off the shelf therapy’. However long-term reconstitution of allogeneic EBV-specific T cells is unlikely, and the need for prolonged culture adds to technical complexity and costs
High doses of unmanipulated EBV-seropositive allogeneic donor lymphocyte infusions have been used successfully to treat EBV-related PTLD. Its use requires less technical expertise than adoptive immunotherapy using EBV-specific T cells. However, donor lymphocytes contain alloreactive T cells that can mediate GvHD. Use of haplo-identical maternal donor lymphocyte infusions reduces the risk of GvHD and can induce remission
EBV's in vitro transforming abilities, its distinctive viral latency, the viral clonality within the malignant cells and results to date from virus targeting therapies implicate EBV as critical to the biology of this disease. The rise in transplantation, including high-risk transplants requiring more intense immunosuppression, coupled with improvements in long-term survival, may result in a rise in the incidence of EBV-related and unrelated PTLD. Future research into pathogenesis is needed, that specifically integrates observations from patient material with novel in vivo and in vitro models of EBV-related PTLD.
We would like to thank Madeleine Kersting for graphic support. Work in the laboratory is supported by the NHMRC (Australia), Cancer Council of Queensland, the Queensland Smart State, the Leukemia Foundation, Atlantic Philanthropies, Research Infrastructure and Support Services, and the Roche Organ Transplantation Research Foundation, and the Australian Centre for Vaccine Development.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.