Cyclosporine and tacrolimus are the two commonest drugs used in maintenance following OLTx and inhibit the phosphatase activity of calcineurin through binding of cyclosporine-cyclophilin and tacrolimus-FKBP12 complexes. This inhibits T-cell activation, but because calcineurin and the nuclear factor activated T-cell pathway are not T-cell specific, calcineurin inhibitors (CNIs) are often associated with significant toxicity[13]. In particular, tacrolimus has high rates of diabetes, while cyclosporin is associated with increased hypertension and dyslipidemia[14]. Furthermore, both drugs are associated with end-stage renal failure that can complicate up to 20% of patients following OLTx[15].

Tacrolimus (> 90%) and cyclosporine (> 50%) are concentrated in erythrocytes, and therefore whole blood is used to measure the therapeutic drug levels[16]. Most centres use an ELISA to measure trough levels of tacrolimus, while large clinical trials of OLTx patients treated with cyclosporine show lower rates of rejection and nephrotoxicity complications with monitoring based on either AUC_0-4 or the concentration 2 h following administration[17-19]. Therefore many units (including our own) perform a level 2 h (C2) following the patient’s morning dose.

Setting a therapeutic target for the CNIs has been difficult with standard protocols generalized to managing large number of recipients, but not specific to each patient’s individual clinical situation[20]. CNIs also have a poor dose-level correlation, an unpredictable level-effect association, individual pharmacokinetic differences, and an unclear level-toxicity relationship[21,22]. Side-effects are seen even with CNI levels below the “therapeutic range”[23]. Further problems arise as the monoclonal antibodies used to detect certain metabolites may not capture all biologically active forms of the CNIs[24,25].

Given the level of drug determined by immunoassay is not correlated with immunosuppressive drug efficacy or the level of immunosuppression[22,26,27] the United States Food and Drug Administration (FDA) has gone so far as to reclassify assays for measuring tacrolimus and cyclosporin blood levels indicating that no suitable therapeutic ranges exist and these tests should not be used alone to adjust drug dosing[28].

CNI dosing is impacted by the variable metabolism of the drugs. Tacrolimus is metabolised by CYP3A enzymes in the small intestine and the enzymatic activity can vary by a factor of 5 between patients[29]. Genetic polymorphisms of CYP3A have shown higher tacrolimus clearance and lower levels in some kidney transplant recipients[30] while attempts to evaluate pharmacodynamics directly through monitoring of CNI biological activity have demonstrated correlation between peak levels of CNIs and residual gene expression (by nuclear factor of activated T-cells), but not clinical events[31].

High-performance liquid chromatography was developed for evaluating four cyclosporine degradation products and two related compounds (CyB and CyG)[32]. Initially developed to test quality control of generic formulations, future studies may consider evaluating whether these could have a closer association with outcomes than the cyclosporin blood level[20].

The CNIs are often used in combination with other immunosuppressants. Steroids and induction agents such as basiliximab (anti-IL2) have no specific monitoring mechanisms apart from side-effects, while the optimal dosing and levels of the mTOR inhibitors remain uncertain.

Even if the biological activity of each individual drug could be accurately determined, this would not provide an objective net biomarker of immune function as the cross-reactive effects of the drugs would remain uncertain. As such, therapeutic drug monitoring may continue to assist clinicians in managing patients, but is unlikely to be the dominant method of future immune system monitoring following OLTx.

One of the major influences on drug dosing and immunosuppression following liver transplantation is the presence of complications. In particular, patients who develop sepsis or malignancy following transplantation often have their immunosuppression empirically reduced. Correspondingly, patients undergoing rejection are treated with increased medication. Clearly this is a crude method of monitoring immunosuppression and the purpose of immune monitoring is to optimise immunosuppression prior to the occurrence of clinical events.

Acute cellular rejection is diagnosed on histology based on the commonly accepted Banff criteria[33]. Sampling graft tissue has the further advantage that it can reveal the local ongoing antidonor immune responses[34] and protocol biopsies provide a more accurate marker of graft function compared to liver biochemistry[13]. Surveillance biopsies of the transplanted organ may represent the gold standard for directly assessing the extent of immune activity within the allograft. However, serial biopsies are invasive and almost impractical outside of a research setting[35].

Although commonly used, the aforementioned tests have significant disadvantages and do not provide an accurate marker of a patient’s immune system following OLTx. As a consequence, clinical events and side-effects remain common causes of morbidity and mortality. Many assays have been developed and evaluated with varying results but are yet to achieve use outside of research settings. In general, these assays can be broadly classified as antigen-specific or non-antigen specific and will be discussed below.

Functional donor specific assays may allow detection of immunological states favouring alloimmune quiescence over reactivity[36]. Functional or cytokine kinetics assays may then be applied to determine preemptively whether immunosuppression dosing should be altered.

Limiting dilution assays (LDA) are an example which can provide more precise quantification of immunity to a given stimulus and allow estimation of frequencies of antigen-specific cells participating in an immune response[37]. It requires recipient peripheral blood mononuclear cells (PBMCs) interacting with donor stimulator cells. This can then be used to determine production of different cytokines in the presence of supernatant cultures such as interferon-gamma, interleukin (IL)-5, IL-4, IL-10, IL-13 or TNF-α present in the well[37]. LDA has been employed to show a highly significant correlation between the donor-specific and third-party stimulated IL-4 and IL-10 produced from recipient PBMCs with stable liver graft function compared with rejectors, independent to level of immunosuppression[38].

The main limiting step is availability of donor cells that can be difficult to obtain from cadaveric transplants unless cells are harvested at time of surgery from the spleen or lymph nodes and cryopreserved for future donor-specific assays[20]. Furthermore, the assays often require substantial laboratory work and may need significant amounts of blood and cells for repeated stimulations/experiments.

Mixed lymphocyte reaction (MLR) assays provide an estimate of the primary in vitro response to the direct recognition of allogenic molecules[37]. Their main value is in assessing tolerance - that is MLR responsiveness in the face of clinically evident donor-specific tolerance.

Studies with ³H-thymidine mixed leukocyte responses (MLR) show that enhanced donor-specific alloreactivity persists longer among children with early rejection and is associated with early and late liver rejection[39,40]. To account for the significant variation that is often seen in donor-specific alloresponses, values are often expressed as a ratio to a third-party response known as the immunoreactivity index. A ratio under 1 suggests low rejection risk[40]. However, this assay is non-antigen specific, requires prolonged stimulation and larger amounts of blood than would be routinely feasible in transplant populations[41].

Further enhancements to MLR include combination of results with carboxyfluorescein diacetate succinimidyl ester (CFSE) labelling by flow cytometry[42]. CFSE is an intracellular fluorescent label that divides equally amongst daughter cells and can be used to study cell division[37]. It measures the proliferative response of recipient lymphocytes after culture or stimulation with donor cells. Unlike many other immune monitoring studies, this has been investigated in an interventional study of 51 adult OLTx recipients. Immunosuppression was increased, decreased or maintained depending on results from the MLR compared with 64 OLTx recipients who had standard of care with empirical based management. This showed trends towards improved rates of rejection and survival, but not sufficient to reach significance (P < 0.05)[42]. A MLR-CFSE assay has also been used to distinguish between rejection on suspicious biopsies[43].

To overcome the issues of prolonged stimulations and blood sample requirements common in MLR assays, Ashokkumar et al[41] evaluated a CD154⁺ (CD40L) T-helper and T-cytotoxic cells MLR as measures of rejection risk[41]. This requires < 24 h of stimulation and only 3 mL of blood. These authors identified pre OLTx CD154⁺ cytotoxic T memory cell responses were associated with significantly increased risk (HR = 7.355, P = 0.02) for rejection. This assay can be ordered as PlexImmune™ (Plexison, Pittsburgh, United States) with results in the United States available 2 d after obtaining blood samples. Only small studies have been published to date with PlexImmune in paediatric liver and small intestinal transplant recipients. The assay requires extraction of PBMCs not only from the recipient but also the donor. In some cases when donor cells have been insufficient or unavailable, “surrogate PBMCs” have been used[41] but their validity is uncertain in a clinical population.

Enzyme-linked immunosorbent spots (ELISPOT) quantifies the frequency of previously activated (memory) T cells that respond to donor antigens by producing a selected cytokine in vitro. Recipient T cells are cultured with donor cells on tissue culture plates coated with a cytokine-specific antibody that is detected using labeled secondary antibodies. Each detected spot represents an effector or memory T cell which has been primed to the stimulating antigens[37].

ELISPOT has been proposed as a surrogate marker of allogenic responsiveness in renal transplantation[44-46]. Pretransplant IFN-γ ELISPOT has been associated with rejection risk following renal transplant[44,45,47] which suggests that IFN-γ-producing cells represent cells that have been sensitized to the graft antigens. Thus providing an ex vivo reflection of the evolving in vivo, donor-reactive immune response which may allow patients without a positive response to reduce or withdraw their immunosuppression[7]. Apart from IFN-γ, granzyme B (GrB) has been studied in a small number of paediatric OLTx recipients but failed to predict the occurrence of rejection[48].

The labor-intensiveness and time-consuming nature of these assays, the need for donor cells, the questionable reliability for stored cells along with some inconsistent correlations with clinical outcomes have prevented their broad acceptance as reliable immune monitoring tools[7,49].

After OLTx, haematopoietic donor cells are transferred with the graft from donor to recipient. These chimeric cells may persist in the recipient and be detectable even years post-transplant[50]. It has been hypothesised that developing chimerism may be desirable after OLTx and potentially associated with tolerance[51]. This could allow immunosuppression to be reduced in patients who have detectable chimerism. However, a meta-analysis has failed to demonstrate a significant association between microchimerism and rejection, but techniques of varying sensitivity were used to measure the degree of chimerism[52]. The value and role of chimerism after liver transplantation remains uncertain, and may also differ depending on the time post-transplant[53].

As immunosuppressive drugs ultimately target T-cell function, it would seem logical that assessing T-cell function would provide a potential biomarker for monitoring immune function after transplantation[54]. ImmuKnow (Cylex Ltd, United States) was developed as a biomarker to guide immunosuppressant dosing following solid organ transplantation and was approved by the United States FDA in 2002. ImmuKnow measures adenosine triphosphate produced after stimulation of T-cells with plant lectin phytohemagglutinin (PHA) mitogen[54]. Whole blood is used to ensure that CNIs are maintained during incubation. After overnight incubation, CD4 cells are selected using paramagnetic particles coated with a monoclonal antibody to CD4[54]. ImmuKnow does not correlate with CD4 cell numbers, and the assay is theorized to provide an independent variable[54].

Studies in OLTx recipients have reported contradictory results for ImmuKnow in predicting acute rejection and infection[55-62]. Most of these studies are retrospective, have limited follow-up, heterogenous in study design, and often include multiple solid organ transplants in the analysis despite immunosuppression protocols and clinical event risks differing substantially amongst different transplant populations.

Further, many of these studies only employ single time point measurements and risk potential bias and the effect of confounders. For example, one study assessing ImmuKnow and infection risk declared lower values in patients who suffer an infection following transplant. However, one of the triggers to run the assay in this study was an event such as fever or raised liver biochemistry[63]. Furthermore, a single result cannot be expected to predict the long-term immune function of the patient. Ideally serial measures, correlated with changes in immunosuppressant dosing, would be needed to adequately assess the immune response post OLTx.

To coincide with the multiple studies demonstrating conflicting results, there have been two opposing meta-analyses published[64,65]. One recent meta-analysis by Ling et al[64] suggests a sensitivity of 0.43 (95%CI: 0.34-0.52) and specificity of 0.75 (95%CI: 0.72-0.78) of ImmuKnow for predicting rejection with a diagnostic odds ratio 1.19 (95%CI: 0.65-2.20). This study incorporated multiple organ transplants and when a sub-analysis of liver transplant patients was conducted, results suggested poor sensitivity but improved specificity (sensitivity 0.11 95%CI: 0.01-0.33, specificity 0.94 95%CI: 0.91-0.95).

A separate meta-analysis in liver transplant recipients identified 4 studies which assessed ImmuKnow for both infection risk and rejection, one further study assessing infection specifically, and a further study examining rejection risk alone. All but one study were retrospective, and in general had small patient numbers with short or undeclared periods of follow-up. In this meta-analysis, the ImmuKnow assay was identified as having a diagnostic odds ratio of 14.7 with sensitivity 83.8% and specificity 75.3% for diagnosing infection. When evaluating rejection, a diagnostic odds ratio of 8.8 (sensitivity 65.6%, specificity 80.4%) was noted alongside significant variation amongst studies included in analysis. In particular, the sensitivity ranged from 9.1%-85.7%[65].

A possible explanation for the perceived poor sensitivity of ImmuKnow in detecting rejection may be that it relies on T cell stimulation with PHA mitogen, which is a non-specific antigen that stimulates the adaptive immune system. With the renewed interest in Toll-like receptors, current evidence suggest that the innate immune system also plays a central role in rejection and allorecognition[66-69]. By only stimulating the adaptive immune system, we postulate that the poor sensitivity may reflect ImmuKnow failing to recognize and therefore measure the contribution made by innate immune mediators to rejection processes.

Clearly there have been issues with several studies that incorporate ImmuKnow. However, the assay is FDA approved and with few other options, the assay is employed in several centres. However, there are often no clear protocols and use varies even amongst individual clinicians in the same centre[35]. A large, formal, multi-centre randomized controlled trial would resolve many questions regarding ImmuKnow in regards to its ability to be an objective biomarker of immune function in OLTx patients.

Productions of cytokines vary amongst individuals, and detecting possible polymorphisms in the responsible genes could help in stratifying patients for risk of clinical outcomes. However, in a meta-analysis studying the impact of cytokine gene polymorphism on graft acceptance in clinical transplantation, the only genetic risk factor associated with acute liver rejection was IL-10 polymorphism at position 1082[70] which is associated with low in vitro production of IL-10[71].

Circulating cytokine levels have the benefit of being reasonably easy to determine. However, analysis of published clinical studies correlating circulating levels with immunological status after liver transplantation are confusing and often contradictory[49]. This probably reflects the multitude of confounding factors that impact this patient population, including surgical stress, the associated ischaemia-reperfusion injury, blood transfusions, hepatic regeneration and infectious complications[72].

Some have evaluated multiple factors such as complement and immunoglobulin levels in an attempt to determine an immune competence score to assist in determining risk of infection[73]. This scoring system assigned two points for each of the following: increased levels of baseline IgG, increased levels of baseline IgA, and decreased levels of pre-OLTx C3. This score was found to have a relative risk of infection of 1.99 (P < 0.001) and would be both relatively cheap and employs pathology tests already available in many labs[73]. However, to our knowledge it has not been validated in larger cohorts and would not take into account the multitude of other factors involved in a patient’s immune function after the transplant operation.

In adult allograft recipients there is evidence that Tregs are involved in transplantation tolerance by directly inhibiting the proliferation of effector T cells. A substantial number of donor Tregs detach from the liver graft during perfusion and continue to migrate into the recipient after OLTx. These suppress the direct pathway alloresponses and are theorized to contribute to chimerism-associated tolerance in vivo in the early stage after transplantation[74].

Lower levels of these regulatory cells have been identified in patients undergoing acute rejection[75,76] while patients completely weaned off immunosuppression demonstrate higher numbers in their grafts and peripheral circulation[77-81]. Despite this, Treg analysis still requires significant laboratory work to isolate PBMCs and perform laboratory analysis and are not currently marketed or used in clinical settings that we are aware of.

Both CD4 and CD8 cells express CD30 after primary alloantigenic stimulation. Although there is some suggestion that soluble CD30 may be a useful marker in kidney transplantation[82,83], studies in adult[84] and paedeatric[85] liver transplantation have failed to reveal a role in predicting rejection outcomes.

The liver allograft can often be maintained after transplantation with low levels of immunosuppression and in some cases be withdrawn completely without histological damage from rejection - defined as operational tolerance (OT)[86]. It is estimated that OT rates after OLTx are as high as 20%-25%[87,88]. It appears that OT recipients have different cellular immunophenotypic or peripheral blood transcriptional profiles compared with healthy volunteers, recipients on immunosuppression or those experiencing rejection[80,86]. Several studies have sought to identify which patients are likely to achieve OT which could then facilitate drug withdrawal in this select group.

Martínez-Llordella et al[89] identified and validated a “tolerant genetic fingerprint” using transcriptional profiling from transplant PBMCs. This identified a modest number of genes capable of identifying tolerant liver recipients with good accuracy. In particular, NK and γδTCR⁺ T cells were the main PBMC subsets associated with tolerance-associated transcriptional patterns.

Although transcriptional profiling of peripheral blood may allow identification of some patients capable of completely weaning off immunosuppression, data directly supporting these assays and their ability to monitor the net immunosuppressive state are yet to be published and not available in clinical settings[20].

In humans, 2 major types of blood dendritic cells have been described[90]. Monocytoid DC (CD11c⁺) can be derived from circulating monocytes in response to granulocyte-macrophage colony-stimulating factor and IL-4 and induce Th1 cell differentiation in vitro and may be specialized for induction of immunity. Plasmacytoid DC (CD123⁺) develop after stimulation with IL-3 and CD125⁺ (CD40L) and promote Th2 responses which can be for induction of tolerance[91]. The ratio of these cells may be important, with flow cytometry demonstrating operationally tolerant patients exhibiting higher incidence of plasmacytoid dendritic cells (theorised to induce tolerance) compared with myeloid dendritic cells[92,93].

In OLTx patients, the trans vivo delayed-type hypersensitivity (DTH) assay has been shown to be valuable in identifying OT recipients[94]. This technique involves transfer of PBMCs plus donor antigen in the footpads of naive, severe combined immunodeficiency mice and measuring for response[94]. This has the advantage of evaluating in vivo cell-mediated allogenic immunity without direct exposure of patients[95]. The logical limitation is the need to have immunodeficient mice available and this makes the assay unfeasible outside research.

Identifying patients who can achieve OT would prove valuable in reducing immunosuppression and related side-effects in these recipients. It would also reduce the ad hoc nature that is sometimes employed to withdraw immunosuppressants following OLTx. However, only a small proportion of patients are likely to have the potential to achieve full operational tolerance and other methods of immune monitoring are therefore needed for the majority of patients.