Operational tolerance: Past lessons and future prospects


  • Josh Levitsky

    Corresponding author
    1. Division of Hepatology and Comprehensive Transplant Center, Northwestern University Feinberg School of Medicine, Chicago, IL
    • Assistant Professor of Medicine and Surgery, Division of Hepatology and Comprehensive Transplant Center, Northwestern University Feinberg School of Medicine, 676 North St. Clair Street, Suite 1900, Chicago, IL 60611
    Search for more papers by this author
    • Telephone: 312-695-9286; FAX: 312-695-0036


Every liver transplant (LT) center has had patients who either self-discontinue immunosuppressive (IS) therapy or are deliberately withdrawn due to a research protocol or clinical concern (ie, lymphoproliferative disorder [LPD], overwhelming infection). This is understandable because maintenance IS therapy, particularly calcineurin inhibitors (CNI), is associated with significant cost, side effects, and considerable long-term morbidity and mortality. Detrimental effects of IS therapy include increased risk of cardiovascular disease, metabolic syndrome, bone loss, opportunistic and community-acquired infections, and malignancy. In fact, LT recipients have among the highest rates of chronic kidney disease and associated mortality among all nonkidney solid organ recipients. This mortality is only ameliorated by undergoing a curative kidney transplant, usurping costs and valuable organ resources. The search for improved treatment algorithms includes trial and error CNI dose minimization, the use of alternative IS agents (antimetabolites, mammalian target of rapamycin [mTOR] inhibitors), or even complete CNI withdrawal. Yet those who are successful in achieving such operational tolerance (no immunosuppression and normal allograft function) are considered lucky. The vast majority of recipients will fail this approach, develop acute rejection or immune-mediated hepatitis, and require resumption of IS therapy. As such, withdrawal of IS following LT is not standard-of-care, leaving clinicians to currently maintain transplant patients on IS therapy for life. Nonetheless, the long-term complications of all IS therapies highlight the need for strategies to promote immunologic or operational tolerance. Clinically applicable biomarker assays signifying the potential for tolerance as well as tolerogenic IS conditioning are invariably needed if systematic, controlled rather than “hit or miss” approaches to withdrawal are considered. This review will provide an overview of the basic mechanisms of tolerance, particularly in relation to LT, data from previous IS withdrawal protocols and biomarker studies in tolerant recipients, and a discussion on the prospect of increasing the clinical feasibility and success of withdrawal. Liver Transpl, 2011. © 2011 AASLD.


APC, antigen-presenting cell; AZA, azathioprine; CD, clusters of differentiation; CNI, calcineurin inhibitor; CyA, cyclosporine A; DDLT, deceased donor liver transplant; FOXP3, forkhead/winged helix transcription factor 3; HCV, hepatitis C virus; HLA, human leukocyte antigen; IFN, interferon; IL, interleukin; IS, immunosuppression; LDLT, live donor liver transplant; LPD, lymphoproliferative disorder; LT, liver transplant; MHC, major histocompatibility complex; mTOR, mammalian target of rapamycin; NK, natural killer; NKT, natural killer T cells; PCD, programmed cell death; PD-1, programmed cell death-1; pDC, plasmacytoid dendritic cell; TAC, tacrolimus; TCR, T cell receptor; TGF-β, transforming growth factor-beta; Th2, T helper 2; Tr1, type 1 T regulatory cell; Treg, T regulatory cell.


Experimental support for transplant tolerance was demonstrated in the early 1950s, when Billingham and Medawar induced tolerance to skin allografts in mice by injection of allogeneic cells in the neonatal period.1-3 Key discoveries from that era led to understanding the role of the thymus in immune system development and self-immune and alloimmune tolerance (central tolerance). They also led to the demonstration of immunoregulation or deletion/anergy of autoreactive/alloreactive cells in the circulation (peripheral tolerance). Figure 1 displays the proposed mechanisms of central and peripheral tolerance. In central thymic tolerance, CD4+, CD8+ cells are initially positively selected to become either CD4+ or CD8+ depending on their TCR interactions with MHC class I (CD8) or II (CD4) molecules on thymic APCs (dendritic cells, macrophages, thymic medullary epithelium). Alternatively, they may undergo PCD if MHC engagement fails. Negative selection follows in which most self-reactive CD4+ or CD8+ cells undergo TCR-induced death if they interact with MHC molecules on APCs carrying self peptides. Alloreactive cells can also undergo negative selection in the thymus during the interaction of antigen-specific T cells with “tolerogenic” dendritic cells (DC2) expressing varying levels of costimulatory molecules.

Figure 1.

Basic mechanisms of tolerance. This figure displays the central and peripheral mechanisms for regulation of autoimmune/alloimmune responses. CD3+ cells that are initially double positive for CD4 and CD8 first undergo positive selection to become either CD4+ (MHC class 2 interaction) or CD8+ (MHC class 1 interaction) followed by deletion or negative selection if autoreactive/alloreactive. However, some CD4+ and CD8+ cells can escape negative selection and become autoreactive/alloreactive Th1 cells in the periphery (lymph nodes, blood, spleen). Mechanisms to regulate these peripherally reactive T cells that can cause autoimmune diseases and transplant rejection are displayed: (1) Immunoregulation from natural (thymic-derived) or induced (in the context of proregulatory cytokines/dendritic cells) Tregs or other regulatory cells; (2) activation-induced cell death; and (3) no activation resulting in programmed cell death.

In peripheral tolerance, autoreactive/alloreactive cells escaping thymic deletion may be deleted or controlled in the circulation by activation-induced cell death, PCD, or the suppressive action of Tregs and cytokines. Alloreactive cells repeatedly stimulated with alloantigen can undergo activation-induced cell death by “death receptors” (Fas) cross-linking Fas ligand, leading to caspase cascade activation and cell apoptosis. This pathway is critical to the inhibition of cell proliferation as well as to the development of tolerance.4-6 Activated cells can also undergo PCD due to the absence of the antigen or activation stimulus. Cell surface receptors, ie, PD-1, and its inducible ligand PD-L1 down-regulate T cell activation. Thus, alloreactive CD4+ and CD8+ cells deficient in PD-1 have enhanced proliferation and avoid anergy.7 This suggests that activation of this pathway is also required for the development of unresponsiveness and tolerance. Finally, certain immature APCs such as plasmacytoid dendritic cells (DC2) are poor presenters of alloantigens on MHC molecules and express low levels of costimulatory molecules (CD80/86, CD40), resulting in diminished signals 1 and 2 necessary for T cell activation.

Pure depletion of alloreactive clones is still unlikely to lead to full immunological tolerance in immunologically mature humans. Active suppression of clonal proliferation by endogenous regulatory cells is presumably required. As such, naturally occurring thymic (CD4+CD25highFOXP3+) and peripherally inducible Tregs (the latter being CD4+CD25 FOXP3 converted to CD25+FOXP3+) promote tolerance to autoantigens and to alloantigens in the context of transplantation.8-10 After TCR engagement, CD4+CD25high Tregs characteristically express an intracellular protein FOXP3,11, 12 which is critical for the maintenance and function of the major Treg subpopulation (CD4+CD25highFOXP3+). The absence of FOXP3 leads to severe autoimmune disease in mice and humans.13-15 FOXP3 blocks the transcription of T cell activation molecules such as IL-2 and the expression of the IL-7 receptor (CD127), the absence of which can distinguish Treg cells (CD127low) from effector and memory T cells (CD127high).16 Following activation, Treg cells are also less sensitive to apoptosis than cytotoxic effector subsets.17 As alluded to above, the CD4+CD25highFOXP3+ cells are not the only putative Tregs, because other populations (CD25 TGF-β–producing Tr1 cells, IL-10–expressing cells, CD8+CD28 suppressors, “regulatory” B cells) have all been characterized as immunoregulatory. These mechanisms involve a number of cellular interactions, regulatory (IL-4, IL-10, TGF-β) cytokines, specialized chemokine receptors, autocrine/paracrine recruitment of more Tregs (“infectious tolerance”), and immature dendritic cells (DC2) promoting alloreactive cell anergy and Treg generation.12, 18

Although having complexity in cell make-up, transplant tolerance is perhaps more simply defined as allospecific immunologic unresponsiveness, either due to deletion of alloreactive T cells (occurring naturally or with immunosuppressive therapy) or by active suppression from allospecific regulatory cells. Table 1 lists the immunological criteria of which some or all appear to be important in transplant tolerance.19, 20 Immunosuppression appears to play an important role in the initial induction and maintenance of tolerogenic mechanisms, particularly early after transplantation when alloreactive and inflammatory signals are at their peak. A notable example of this is the potential association of tolerance with lymphodepletional induction therapy,21-24 either by depletion of alloreactive clones or augmentation of Tregs. This is in contrast with antitolerogenic mechanisms seen with ischemia/reperfusion injury and subsequent Toll-like receptor activation of APCs.25-29 Donor/recipient hematopoietic microchimerism has been demonstrated in long-term liver and kidney transplant survivors and postulated to result in reciprocal donor/recipient cell exhaustion (ie, absence of both rejection and graft-versus-host disease).19, 30-34 Controversial, however, is the actual requirement of chimerism for tolerance because this may only be a transient phenomenon.35, 36 Also contentious is the role of FOXP3+ expression in human organ transplant recipients, because FOXP3+ cells are found with rejecting organ transplants. More evidence is needed on the actual function of donor-specific and nonspecific FOXP3+ Tregs coming from and migrating to various tissue compartments (blood, lymphoid organs, bone marrow, and allograft).

Table 1. Immunological Criteria for Transplant Tolerance
Early immune quiescence (ablation of alloreactive effector T cells and inflammatory signals) induced by potent immunosuppressive therapy and minimization of ischemia/reperfusion injury
Inactivation of immunoreactive clones by an “exhaustion” process using reduced dose maintenance or tolerogenic immunosuppression
State of regulatory (IL-10, TGF-β) versus inflammatory (IL-6, IFN-γ) cytokine production and gene/protein expression
The establishment of a myeloid and lymphoid chimeric state by donor-derived, blood-forming elements
Expansion of regulatory T cells (CD4+CD25high FOXP3+) and immature dendritic cells to favor immunoregulation over reactivity
Immunogenetic similarity


The liver itself represents a unique “window” to the immune system and is the most immunoregulatory solid organ that is transplanted (Fig. 2). It contains a high number of extramedullary hematopoietic cells, a large mass of nonhematopoietic cells (hepatocytes, stellate cells, endothelial cells), and secretes a variety of proteins (ie, HLA-G) and cytokines with immunoregulatory effects.37-39 The abundance of resident immunocytes and APCs appear to be regulatory in nature and protective of graft injury and rejection, although in some cases can lead to early graft-versus-host disease.37 These cells might even engulf (“suicidal emperipolesis”) recipient allospecific cytotoxic T cells as one of the pathways of liver immunoregulation.40 Donor-specific immunoregulatory effects, clonal deletion of alloreactive immunocytes, dilution or inhibition of alloantibody, and mixed donor-recipient hematopoietic microchimerism are also putative components of liver immunoregulation.41-48 Moreover, the liver graft itself may itself be considered a source of persistent tolerogen in contrast to being a target of immune destruction.

Figure 2.

Putative components of liver immunoregulation. This figure illustrates clinical and immunologic evidence supporting liver transplant immunoregulation. The significant contribution from resident regulatory cells and cytokines is likely a major factor leading to low rejection rates and IS requirements.

Consistent with these mechanisms and distinct from other solid organ recipients, LT recipients demonstrate a greater ability to withstand lower IS doses and have a propensity toward tolerance over alloreactivity. This is evident clinically by the insignificance of HLA matching, minimal relevance of a single rejection episode (except in HCV-positive recipients), the low incidence of acute/chronic rejection, the reduced degree of induction/maintenance IS required, and the immunological protection conferred by the liver on other organs in combined transplants (liver–kidney, liver–intestine). That being said, these basic mechanisms and clinical phenomena cannot be generalized to all LT recipients, and the reasons why minimized or no IS is achievable in only some recipients have not been sufficiently elucidated.


Although immunological and clinical evidence exist for the development of LT tolerance, complete withdrawal of IS (mainly CNI) maintenance therapy is only successful in ∼20% of recipients in prospective studies to date (Table 2; ≥5 patients/study).35, 49-61 This low percentage is likely due to known CNI mechanisms inhibiting immunoregulation and the lack of available, well-defined immune monitoring to detect immunoregulation or unresponsive states. This inability to immunologically predict or achieve successful withdrawal has compelled clinicians to maintain long-term CNIs and other IS at recommended doses, despite the toxicities and morbidity. However, the failures and successes from the previous studies present useful lessons on specific patient characteristics (primary liver disease, time from transplant, type of transplant) to guide more effective approaches to withdrawal.

Table 2. Elective Withdrawal Studies
Center (No. of Patients)Adult or PediatricDDLT or LDLTBaseline ISYears from LT to TaperingTolerantFailure*
  • *

    Either due to rejection, immune-mediated hepatitis, noncompliance, resumption of immunosuppression, disease recurrence, or other. The remaining patients were deemed “weaning in progress” in all studies.

  • Randomized controlled trial of ursodeoxycholic acid given at 15 mg/kg/day versus placebo in withdrawing patients; 3 patients developed autoimmune hepatitis recurrence after withdrawal.

  • 45 received donor bone marrow cell infusions; 59 did not.

Pittsburgh (n = 95)BothDDLTTAC or CyA + AZAMean, 8.4 ± 4.718 (18.9%)40 (42.1%)
London (n = 18)AdultDDLTCyA, AZA, prednisoloneMedian, 7 (5-11)5 (27.7%)13 (72.2%)
Kyoto (n = 115)PediatricLDLTTAC>249 (42.6%)20 (17.4%)
Murcia (n= 9)AdultDDLTCyAMedian, 5.1 (2-9)3 (33.3%)6 (66.6%)
Rome (n = 34, only HCV)AdultDDLTCyAMean, 5.3 ± 1.78 (23.5%)26 (76.5%)
New Orleans (n = 18)AdultDDLTTAC>0.51 (5.6%)17 (94.4%)
Winnipeg (n = 26)AdultDDLTCyA + AZA or prednisoloneMean, 4.3 ± 1.18 (30.8%)18 (69.2%)
Miami (n = 104)AdultDDLTTAC or CyAMedian, 4 (3.6-4.6)23 (22.1%)81 (61.5%)
Barcelona (n = 102)AdultDDLTTAC or CyAMedian, 7.940 (77.9%)62 (60.0%)

The vast majority of withdrawal studies have only focused on the nonimmune, non–viral infected population. Most studies excluded patients with a history of immune-mediated diseases, ie, autoimmune hepatitis, primary biliary cirrhosis, and primary sclerosing cholangitis, or with a more significant or recent history of rejection. Yet, the negative impact of chronic IS therapy in this population may be higher, because many are empirically treated with long-term double or triple therapy to avoid rejection. Although it makes clinical sense to not include these “immunoreactive” patients in tolerance trials, it might be beneficial to determine clinical phenotypes, graft histology, or biomarkers associated with a greater success of IS minimization (proper tolerance) rather than complete withdrawal. Approaches involving more aggressive immune manipulation (myeloablation plus stem cell therapy) have not been studied in this population but warrant consideration.

Only one study has examined the effect of IS withdrawal in HCV-positive recipients, an important group to consider given the known negative impact of IS on HCV recurrence and fibrosis progression.57 This study demonstrated improvement in fibrosis after withdrawal, similar to that seen with successful post-LT IFN therapy.62, 63 However, this preliminary study has not been replicated, and a follow-up study almost 3 years later did not show histological differences between HCV-positive transplant recipients in whom IS was withdrawn and those in whom it was maintained.64 In addition, the development of graft rejection associated with withdrawal of IS in HCV-positive patients may have an opposite effect on HCV recurrence, fibrosis progression, and long-term graft survival. To address these concerns, larger prospective studies including HCV recipients in withdrawal in conjunction with specific immunological monitoring are currently being conducted.65, 66

The length of time between transplantation and IS withdrawal might also affect the success and clinical benefit of withdrawal. Many of the studies in Table 2 excluded patients who were within 3 years of LT, so it is difficult to compare the pros and cons of early versus late withdrawal. Newer trials in IS withdrawal have enrolled stable LT recipients for withdrawal as early as one year post-LT with or without induction therapy,65, 67 with one study showing high rates of rejection following a protocol of antithymocyte globulin induction and early TAC weaning.67 A more recent study, not yet formally published, demonstrated greater success when patients were withdrawn late (>5 years) after LT.49 Although IS withdrawal may be intuitively more successful when done later after LT, the potential for IS adverse effects is greater the longer patients are maintained on IS. Overall, at this juncture, there does not seem to be clear linear correlation between withdrawal success and time from transplantation. Thus, strategies designed to “tolerize” patients and withdraw IS earlier after LT may have more beneficial impact on IS-related morbidity.

In regard to type of transplant, ie, LDLT versus DDLT or pediatric versus adult, the Kyoto group demonstrated a high IS withdrawal success rate (42.6%) fairly early after pediatric LDLT.55, 56 Other case series have reported IS withdrawal in recipients of simultaneous or staggered donor hematopoietic cell transplantation and LDLT.51, 68-71 The success of these approaches may be in part due to greater HLA similarities between the related donor/recipient pairs or the avoidance of significant ischemia/reperfusion injury by the nature of the procedure. However, other than the Kyoto report, no firm clinical data are available to suggest that LDLT recipients require less IS or can be withdrawn more easily compared to DDLT. In addition, no head-to-head trials comparing IS withdrawal in LDLT versus DDLT, with or without stem cell therapy, have been or are currently being performed, but certainly are of great interest and potential.

Finally, the issue of IS withdrawal in children is important given the negative, life-long impact of IS therapy in this population. Only the Pittsburgh and Kyoto groups have a large enough prospective experience in children, succeeding in achieving tolerance in 22 (34%) and 49 (43%) recipients, respectively.54, 55 Favorable factors for success include longer time from transplantation, achievement of monotherapy, transplantation for nonimmune diseases, the presence of LPD, and younger age.72, 73 Of unclear importance is the presence of fibrosis and decreased bile duct size on postwithdrawal protocol biopsies of “operationally tolerant” pediatric recipients.60 More data on long-term histology of withdrawn patients will be available soon to clarify the incidence, etiology, and significance of such graft abnormalities as well as other clinical outcomes.74


There have been reports of successful IS discontinuation in the setting of LPD53, 75 or by self-withdrawal without the knowledge of the center. Other than in research protocols or life-threatening infections, LPD is perhaps the only situation in which there is general clinical consensus to intentionally taper or withdraw IS, usually on a temporary basis until the completion of chemotherapy. The self-withdrawal group, although few and far between, represents fortunate patients who “escaped” rejection. In turn, transplant researchers in some cases have obtained samples from these spontaneously or research protocol tolerant patients and their “tolerant phenotype” has allowed testing of in vitro bioassays as putative in vivo markers of tolerance. The major candidates are blood immunophenotypic assays (CD4+CD25highFOXP3+ cells, and dendritic and Vδ1/Vδ2 cell ratios), cytokine gene profiles (NK, γδ T cell, CD8+ receptor genes) and genomic microarrays, although donor-specific assessments and immune characterization of allograft tissue need to be more robustly developed as specific identifiers of tolerance (Table 3).

Table 3. Candidate Liver Transplant Tolerance Assays
  1. The more definitive assays have been validated in operationally tolerant liver transplant recipients. The less definitive assays are either inconclusive or inadequately studied in clinical tolerance protocols.

More Definitive
 • Peripheral blood immunophenotyping
   – CD4+CD25highFOXP3+ cells
   – DC2:DC1 ratio
   – γδ T cells (Vδ1/Vδ2 ratio)
 • Genomic signatures
   – NK cell, γδ T cell, CD8+ cell receptors
   – Cytokine gene polymorphisms (TNF-α, IL-10)
 • Soluble HLA-G
Less Definitive
 • Allograft immunophenotyping
   – Immunohistochemical staining and in vitro culture for Treg:Teff ratio
   – FOXP3 mRNA
 • Donor-specific approaches
   – Mixed lymphocyte reaction (proliferation, CFSE labeling, Treg MLR)
   – Cell-mediated lymphotoxicity
   – ELISPOT (Th1 and Th2 cytokines)
   – Delayed type hypersensitivity (Trans-vivo)
   – Donor-specific antibodies and HLA typing
 • Detection of hematopoietic chimerism

Operationally tolerant recipients appear to have different cellular immunophenotypic profiles compared to healthy volunteers and recipients maintaining IS or experiencing rejection. Studies have shown that operationally tolerant pediatric and adult LT recipients have significantly higher peripheral blood CD4+CD25high T cells and γδ T cells (Vδ1/Vδ2 ratio) compared to nontolerant recipients or healthy individuals.61, 76 In a subset, the suppressive properties of the isolated CD4+CD25+ cells were donor-antigen specific.60 Blood FOXP3 transcripts in conjunction with CD4+CD25high T cells were observed in higher frequency in LT recipients who underwent successful withdrawal versus those who developed rejection.77 Other cell populations observed to be higher in withdrawing and fully tolerant recipients include plasmacytoid “regulatory” dendritic cells (DC2 or CD11cCD123high), particularly the ratio of DC2:DC1 (DC1 or CD11c+CD123−/lo+) and B (CD19+) cells, the latter also seen in tolerant kidney transplant recipients.61, 78-80 Interestingly, in the allograft specifically, FOXP3 expression and CD4+CD25high T cells increase in both tolerant and rejecting liver81, 83 and other organ recipients.84, 85 The presence of graft FOXP3+ cells in rejection likely represents either activated alloreactive T cells or regulatory T cells homing to the graft to control immune responses. Thus, even with available histological analysis and a “window” to the graft itself, immunohistochemical characterization of graft immunocytes may be less useful and predictive than peripheral blood analyses that have shown more consistent results.

There have also been reports of cytokine, genomic, and HLA signatures present in tolerant patients that might predict the ability to achieve tolerance prior to weaning.49, 76, 86-92 An early study demonstrated that pediatric patients on minimal or no IS had low tumor necrosis factor-α and high IL-10 gene polymorphism profiles compared to control patients on maintenance IS.93 More recently, gene expression profiling of operationally tolerant liver recipients demonstrated a unique signature involving receptors for γδ T cells and NK cells, as well as proteins involved in cell proliferation arrest.76 This correlated with increased numbers of circulating putative Tregs (CD4+CD25+) and γδ T cells (mainly the Vδ1+ subset). The same group found similar gene (NK, γδ, and CD8+ cells) and immunophenotypic (CD4+CD25+, γδ T cells of the Vδ1+ subset) signatures in LT recipients who were successfully withdrawn from IS in an actual weaning protocol compared to healthy controls and recipients who rejected with weaning.86 This signature appears to be specific to tolerant LT recipients because a different B cell–related signature has been demonstrated in tolerant kidney recipients.94, 80 Whether microchimerism plays an active role in immune quiescence may depend on the characterization of the chimeric immunocyte subsets and perhaps such genomic and transcriptional predilections. Finally, levels of soluble HLA-G have been shown to be significantly higher in tolerant pediatric recipients compared to those with rejection or stable on IS therapy.95 Although prospective studies are currently ongoing,49 sequential data are not yet published to directly support the ability of these “tolerance” assays to predict successful withdrawal, nor are they yet commercially available in clinical settings.


Currently, few if any transplant clinicians would consider IS withdrawal a feasible option. Without predictive tools or clinical guidance, the risks currently outweigh the small potential for success. The key for the future is determining which clinical and immunological characteristics identify the population most likely to succeed, such that withdrawal would only be considered in such a suitable group. Conversely, those with unfavorable characteristics and bioassays would not be considered for IS withdrawal and thus avoid the accompanying risk of rejection and graft loss. Fortunately, we are closer to having predictive immunological and genetic assays (Table 3) that might be clinically serviceable. In addition, the tolerogenic profiles of IS agents, well described in vitro, need to be clarified in vivo to determine optimal IS protocols that facilitate tolerance. Although controversial, the tolerogenic potential may be greatest with certain induction (thymoglobulin, alemtuzumab) and maintenance (mTOR inhibitors) agents and the least with CNIs and IL-2 receptor (anti-CD25) agents. The next section will focus on how different IS therapies and bioassays might allow for expansion of the number of tolerant recipients.

Induction Immunosuppression Selection

A more aggressive approach would be to use up-front lymphodepleting induction therapy to eliminate early detrimental immune signals and promote clonal deletion of effector cells. Ischemia/reperfusion injury elaborates proinflammatory innate cytokines (tumor necrosis factor-α) that activate Toll-like receptors, promotes alloantigen presentation by APCs, and inhibits immunoregulation. Reduction in ischemia/reperfusion injury (ie, LDLT) in combination with induction therapy might putatively alter the balance toward immunoregulation over alloreactivity. Studies have demonstrated the possible regulatory effects of antithymocyte globulin and the anti-CD52 antibody alemtuzumab in deleting alloreactive effector cells while preserving regulatory cells and their components.96

However, the limited existing data suggest that induction approaches may not increase the success of IS withdrawal and perhaps rather favor a state in which minimized IS doses can be achieved.20, 58 More recently, an induction approach (antithymocyte globulin; ATG-Fresenius) in conjunction with early IS (TAC) withdrawal was associated with a high rate of rejection and actually did not allow for significant minimization of TAC.67 Also of concern is the presence of memory effector cells that are more resistant to lymphodepletional therapy. If possible, elimination of donor-specific memory cells and preservation of nonspecific memory responses might promote tolerance and maintain important immune surveillance against pathogens and tumor cells, respectively. It is also uncertain if costimulatory blockade molecules (cytotoxic T lymphocyte antigen 4 [CTLA-4] immunoglobulin) or IL-2 receptor inhibitors being developed or already in practice have pro- or antitolerogenic effects, although negative effects are more likely because Tregs express CTLA-4 and respond to IL-2 for proliferation. In summary, the putative immunoregulatory properties of lymphodepletional or other induction approaches do not, as of yet, appear to translate clinically in facilitating IS withdrawal in LT recipients. Randomized trials comparing induction versus noninduction tolerance protocols in LT recipients are needed to clarify the overall risks and benefits of these approaches.

Maintenance Immunosuppression Selection

Previous nonclinical studies have demonstrated differences in immunoregulatory properties of maintenance IS agents. The CNI agents inhibit the calcineurin-driven pathways of IL-2 and IFN-γ transcription resulting in inhibition of T cell activation. However, these same CNI mechanisms are counter-regulatory, because IL-2 is critical for FOXP3 expression and the survival and proliferation of CD4+CD25highFOXP3+ Tregs.97-99 In addition, CNIs do not affect the maturation of dendritic cells, which present alloantigen and produce costimulatory molecules that activate effectors and inhibit Treg generation.97, 98 Conversely, whereas not yet clinically established, the mTOR inhibitors (rapamycin, everolimus) and possibly the antimetabolites (mycophenolic acid) suppress alloreactive T cells but also enhance Treg generation in vitro. Unlike CNIs, these agents do not inhibit initial IL-2 transcription that is important for the maintenance and function of Tregs. When administered alone or with costimulatory blockade/IL-10 in vitro, they inhibit the maturation and function of IFN-γ–producing T helper 1 (Th1) cells and dendritic cells, and increase the percentage of CD4+CD25high FOXP3+ Tregs and CD8+CD28 suppressor cells.100-102 Rapamycin also appears to promote regulatory costimulatory molecules.103

In the clinical setting, limited data have shown that liver and kidney transplant recipients treated with mTOR agents have higher percentages of phenotypic Tregs versus those recipients on CNI therapy.104-107 As such, one of the possible reasons for the low success rates in prior weaning studies could be that the majority of patients were withdrawn directly from CNI therapy that may have inhibited regulatory mechanisms. If conversion from CNI to putative “tolerogenic” IS therapies such as rapamycin is found to enhance the generation of Tregs and other tolerance biomarkers, this might be an interventional step to facilitate the ability to increase the number of tolerant patients. However, without supportive clinical withdrawal data, this approach is purely theoretical and should be rigorously tested in prospective fashion against the standard agents. In addition, it is not known if these IS-influenced Tregs are functionally regulatory in vivo or regulate donor-specific alloresponses in the peripheral blood as well as the formative (bone marrow) and target (liver) organs.

Use of Biomarkers

The ultimate goal would be to demonstrate persistent recipient hyporesponsiveness or unresponsiveness to the donor. This may now be more feasible with donor-specific immune monitoring assays90, 91 to more clearly detect states favoring alloimmune quiescence over reactivity, ie, whether they are due to partial clonal deletion/exhaustion19, 20 or an active regulatory process.108 These functional assays (mixed lymphocyte reactions, ELISPOT, Trans-vivo delayed-type hypersensitivity, cell-mediated lympholysis) might then be used to monitor patients before, during, and after IS minimization and withdrawal, allowing continuation of withdrawal in those with favorable donor-specific responses and avoiding such interventions in those with immunoreactivity. In the absence of donor-specific assays, nonspecific genomic (NK, γδ T cell genes), immunophenotypic (CD4+CD25highFOXP3+, γδ [Vδ1+] T cells), and HLA (soluble HLA-G) signatures described previously in this review (Table 3) appear to be ready for testing as tolerance predictors in prospective studies. Ideally, along with standard liver histology, they could be used to determine suitable and unsuitable candidates for weaning. For example, out of 100 patients, 30 might speculatively have the genomic and immunophenotypic tolerance signatures prior to weaning, of which perhaps 20 will wean successfully. This still results in the same 20% success rate seen in prior studies, although the rate of rejection has been reduced from 80% to 10% by only weaning the 30 patients with the signature. In addition, the use of tolerogenic IS approaches in combination with predictive signatures to guide decision-making could eventually alter the balance enough to make weaning feasible and successful on an even greater percentage of LT recipients.

Stem Cell Transplantation

A riskier approach would be to use partially or nonmyeloablative therapies (chemotherapeutic agents, total lymphoid irradiation) followed by donor CD34+ or mesenchymal stem cell infusions to induce tolerance.51, 68 The concept here is that the induction of mixed donor/recipient chimerism might provide a new T cell repertoire tolerant to both recipient and donor and facilitate earlier removal of IS therapy. However, precursor alloreactive T cells would need to be aggressively depleted for tolerance induction, particularly in MHC-mismatched allografts.109 Such higher intensity conditioning protocols geared toward profound T cell elimination would carry high risks of infectious and malignant (LPD) complications and may only be suitable in less sick recipients (those with low MELD scores) or LDLT pairs with better HLA matching. Whether these strategies are ultimately required for donor-specific tolerance to develop in liver recipients is not known, although given the toxicity, they may be more intuitive for organ transplants (kidney, pancreas, cardiac) with a higher barrier to tolerance.36


The limitations of past experiences have indeed provided valuable lessons in considering future approaches to withdrawal. First, it is likely too restrictive to only consider patients without viral or immune diseases for withdrawal, because these populations may arguably derive even greater clinical benefits. Hence, carefully designed, disease-specific withdrawal approaches incorporating clinical, immunological, and histological monitoring should support inclusion of these patients in such trials. Second, the “requirement” for a prolonged quiescent time from transplantation to withdrawal is contentious. Comparative trials of early versus late withdrawal that incorporate not only overall success rates but also clinical risks and benefits are needed. Similarly, comparative LDLT versus DDLT withdrawal trials are required to better establish the most optimal approaches and patient selection. Third, trials need to not only include patients at highest risk of IS complications (those with HCC, HCV, chronic kidney disease), but also children and young adults who are subject to longer lifetime IS exposures.

Finally, the true benefit of IS withdrawal on health outcomes, physical and mental quality of life, and costs has not been fully characterized, likely due to the low percentages who have actually reached this endpoint or followed in withdrawal studies.52 Some studies have preliminarily suggested that IS withdrawal may improve CNI side effects, such as kidney function, hyperlipidemia, hypertension, and diabetes, and quality of life.64, 110 Yet, a limitation of all withdrawal studies is the absence of prospectively followed, IS-maintained patients as control cohorts. In addition, IS minimization strategies such as once daily or more intermittent dosing might speculatively be more beneficial to the LT population than full withdrawal and provide the most optimal approach. As such, long-term outcomes and IS-related effects, both specific (HCV/HCC recurrence) and general (renal, metabolic, infection, malignancy, cardiovascular), need to be prospectively compared in those patients who are successfully withdrawn versus those who are maintained (no attempt at withdrawal; standard or minimized doses) to better characterize the risk/benefit pendulum. These approaches are likely to be more useful in understanding the true clinical benefits of withdrawal rather than comparing such outcomes in tolerant versus intolerant (failed withdrawal; higher risk) recipients.64, 110

In conclusion, rather than identifying such “lucky” tolerant patients ex post facto, prospective approaches incorporating patient characteristics, disease states, tolerogenic IS selection, and immune monitoring should increase the feasibility of achieving tolerance and establish it as a legitimate option for select patients. The era of “lifetime IS therapy for all” needs to and can move into a new era of “personalized IS strategies”, with a significant percentage of recipients undergoing weaning and others being maintained, all based on precisely determined immunological traits, therapies, and biomarkers.