Tacrolimus Trough Levels, Rejection and Renal Impairment in Liver Transplantation: A Systematic Review and Meta-Analysis

Authors

  • M. Rodríguez-Perálvarez,

    1. The Royal Free Sheila Sherlock Liver Centre and University Department of Surgery. UCL, and Royal Free Hospital. Pond Street, London, UK
    Search for more papers by this author
  • G. Germani,

    1. The Royal Free Sheila Sherlock Liver Centre and University Department of Surgery. UCL, and Royal Free Hospital. Pond Street, London, UK
    Search for more papers by this author
  • T. Darius,

    1. Starzl Unit of Abdominal Transplantation, Cliniques Universitaires Saint Luc, Université Catholique de Louvain (UCL), Brussels, Belgium
    Search for more papers by this author
  • J. Lerut,

    1. Starzl Unit of Abdominal Transplantation, Cliniques Universitaires Saint Luc, Université Catholique de Louvain (UCL), Brussels, Belgium
    Search for more papers by this author
  • E. Tsochatzis,

    1. The Royal Free Sheila Sherlock Liver Centre and University Department of Surgery. UCL, and Royal Free Hospital. Pond Street, London, UK
    Search for more papers by this author
  • A. K. Burroughs

    Corresponding author
    1. The Royal Free Sheila Sherlock Liver Centre and University Department of Surgery. UCL, and Royal Free Hospital. Pond Street, London, UK
    Search for more papers by this author

Andrew K. Burroughs, andrew.burroughs@nhs.net

Abstract

We hypothesized that current trough concentrations of tacrolimus after liver transplantation are set too high, considering that clinical consequences of rejection are not severe while side effects are increased. We systematically reviewed 64 studies (32 randomized controlled trials and 32 observational studies) to determine how lower tacrolimus trough concentrations than currently recommended affect acute rejection rates and renal impairment. Among randomized trials the mean of tacrolimus trough concentration during the first month was positively correlated with renal impairment within 1 year (r = 0.73; p = 0.003), but not with acute rejection, either defined using protocol biopsies (r =−0.37; p = 0.32) or not (r = 0.11; p = 0.49). A meta-analysis of randomized trials directly comparing tacrolimus trough concentrations (five trials for acute rejection [n = 957] and two trials for renal impairment [n = 712]) showed that “reduced tacrolimus” trough concentrations (<10 ng/mL) within the first month after liver transplantation were associated with less renal impairment at 1 year (RR = 0.51 [0.38–0.69]), with no significant influence on acute rejection (RR = 0.92 [0.65–1.31]) compared to “conventional tacrolimus” trough levels (>10 ng/mL). Lower trough concentrations of tacrolimus (6–10 ng/mL during the first month) would be more appropriate after liver transplantation. Regulatory authorities and the pharmaceutical industry should allow changes of regulatory drug information.

Abbreviations: 
ACR

acute cellular rejection

LT

liver transplantation

TC

trough concentration

RCTs

randomized controlled trials

HCV

Hepatitis C virus

IQR

interquartile range

ROC

receiver operating characteristic

Introduction

The incidence of acute cellular rejection (ACR) after liver transplantation (LT) has decreased since tacrolimus was introduced (1,2). Tacrolimus is used in most transplant units as it results in less ACR and better survival than cyclosporine (3). Significant ACR, which includes moderate–severe histological grades, is most frequent within 2 weeks after LT and decreases progressively thereafter. In LT the risk of ACR is similar to other solid-organ transplants but it has less clinical relevance (4). Moreover the response to specific therapy (i.e. corticosteroids boluses) is satisfactory in most cases (5), thus evolution to chronic rejection and graft loss is an infrequent event (<5%) (6,7).

Despite this the conventional and regulatory targets for tacrolimus trough concentrations (TC) used in immunosuppression trials in LT are high (i.e. 10–15 ng/mL during the first 4–6 weeks and 5–10 ng/mL thereafter), since these are transposed from kidney transplantation (8). Adverse effects of immunosuppression are also similar being early exposure to tacrolimus especially associated with long-term chronic renal dysfunction (9,10). The increasing need to use older donors, donation after cardiac death and the more frequent renal dysfunction in recipients, has increased renal impairment after LT (11). This emphasizes the need to hallmark the early transplant period, in which changes in immunosuppression may lead to long-term gains.

Nevertheless calcineurin inhibitors are critical in preventing ACR and graft loss; clinical trials testing their complete avoidance were stopped prematurely due to increased ACR (12,13). The main focus in immunosuppression still is optimizing tacrolimus dosing, especially within the first weeks after transplantation, to balance prevention of moderate–severe ACR versus adverse effects, particularly on the kidney. However although the relationship between tacrolimus TC and toxicity is well established (9,14–16), the relationship between TC and ACR is controversial (14,15,17,18). The latter is contributed by the heterogeneity in how ACR is diagnosed: most studies do not use protocol biopsies, but use “clinical suspicion” based on liver function tests to perform a biopsy, misdiagnosing 32% of patients (7,19).

Indeed in a scenario characterized by less rejection and worse renal outcomes (9), LT recipients are overimmunosuppressed (20) (Figure 1). Therefore, it should be possible to decrease target TC of tacrolimus. In this review we evaluate the evidence related to tacrolimus TC, ACR and renal dysfunction following adult LT, and draw conclusions on what TC should be, so as to influence current clinical practice and future trial design.

Figure 1.

Immunosuppression in liver transplantation. Currently the consequences of rejection are much reduced while toxicity and complications derived from immunosuppression are increased.

Materials and Methods

Literature search, study selection and data extraction

A search on MEDLINE, Cochrane Controlled Trial Register (CENTRAL), EMBASE and Science Citation Index databases was done from January 2002 to January 2012, to assess the more current usage of tacrolimus. We identified studies using these key words: “liver transplantation”, “tacrolimus”, “trough concentration”, “rejection” and “renal impairment”, also using equivalent free-text terms, and no language restrictions. The search resulted in 204 records which were categorized and screened independently by MRP and GG (differences resolved by AKB). Duplicate records, reviews and not related articles were removed resulting in 78 eligible studies including 46 randomized controlled trials (RCTs). Another 14 RCT were excluded because randomization occurred late after LT, or concerned pediatric population, or were preliminary reports not providing enough data about renal impairment, rejection or tacrolimus exposure. We finally selected 64 articles comprising 32 RCT and 32 observational studies (Figure 2).

Figure 2.

Flow diagram illustrating the search strategy used according to PRISMA (preferred reporting items for systematic reviews and meta-analyses) statement ( 62 ).

The following data were extracted from each study independently by MRP and ET using a predefined form: year of publication, number of patients, immunosuppression regimen, tacrolimus TC target, mean tacrolimus TC within the first month after LT, ACR rates, renal impairment rates and methodological quality. Evaluated outcomes were biopsy-proven ACR after clinical suspicion or by protocol biopsies, renal impairment at 3–6 months and renal impairment at 1 year after LT.

Methodological quality

Meta-analysis was appropriate for RCT in which tacrolimus with conventional and reduced TC (>10 ng/mL and <10 ng/mL, respectively) were compared in opposite arms. The methodological quality was defined as the control of bias in the treatment comparison. The randomization methods were classified as the primary measure of bias control (21), being evaluated by the allocation sequence generation (classified as adequate if based on a table of random numbers, computer-generated random numbers or similar) and allocation concealment (classified as adequate if based on central randomization, identically appearing coded drug containers, serially numbered opaque sealed envelopes or similar). We also extracted blinding, the risk of attrition bias and whether the primary outcome measure was defined and reported. MRP and ET independently evaluated the methodological quality (any differences resolved by AKB).

Statistical analysis

The analyses were performed using RevMan version 5.1 (Nordic Cochrane Centre, Copenhagen, Denmark). Meta-analyses were performed to allow intention-to-treat analyses, using random effects models due to expected clinical heterogeneity. The number of events and number of patients in each intervention arm were used to calculate relative risks (RR) and 95% confident intervals (CI). I2 values were indicators of the degree of intertrial heterogeneity. The possibility of publication bias was explored with funnel plots (supplementary Figure 1).

Results

Pharmacokinetics of tacrolimus in liver transplant recipients

A small fraction of tacrolimus binds to leukocytes (0.61 ± 0.4%) which are the therapeutic target (22), leading to a narrow therapeutic index, coupled with a long-half life (43 h) and slow renal clearance. Thus monitoring blood TC is necessary to minimize overdosage. In LT the variability of tacrolimus dosage is increased due to changes in blood count, protein concentrations and albumin supplementation (23). Hepatitis C virus (HCV) decreases hepatic CYP3A4 activity, requiring lower doses of tacrolimus over time (24).

Current recommendations

The currently recommended tacrolimus TC after LT come from RCTs (Table 1), in which the stipulated targets for TC were transposed from kidney transplantation (8), and reflect the registration details filed by manufacturers. According to these, the conventional initial tacrolimus doses are adjusted to maintain TC of 10–15 ng/mL during the first 4–6 weeks with a progressive reduction thereafter, achieving 5–10 ng/mL in the long term, with minimal variations between studies (see Table 1). With this regimen adverse effects are common, leading to dose reduction or stopping tacrolimus, which explains the discordance between tacrolimus TC aimed for and actually achieved in some studies (25–27). Indeed in a recent study (28) a target TC of >12 ng/mL was used within the “standard therapy” group during the first 6 weeks after LT: 45% (83/183) of patients needed to stop, mostly due to adverse effects. In contrast a clinical trial reported within the same period, aiming at tacrolimus TC 6–8 ng/mL, only had one withdrawal (0.64%) due to toxicity (29).

Table 1.  Randomized controlled trials comparing different immunosuppressive protocols using tacrolimus in liver transplantation between 2002 and 2012. Only comparison arms using tacrolimus are described
  1. ACR = acute cellular rejection; ATG = antithymocyte globulin; DAC = daclizumab; MMF = mycophenolate; TAC = tacrolimus; STD = steroids; NS = not significant.

  2. *Tacrolimus trough concentrations were measured in whole-blood samples in all studies. The method was tandem mass spectroscopy in Fischer (63) and Nashan (32). Immunoassay was used by Manousou (36), Lerut (29) and Shenoy (67). The remaining randomized trials did not mention the methodology used.

inline image
inline image
inline image
inline image
inline image
inline image

In current clinical practice, most centers target lower tacrolimus TC, often combined with other agents. Indeed TC in RCT of “tacrolimus sparing” are usually lower, but have added agents in the “mistaken” premise to compensate for tacrolimus reduction, thus favoring overimmunosuppression, and which may not be necessary. In the literature a significant difference exists between TC targeted in RCTs (Table 1) during the first 4 weeks (>10–12 ng/mL) compared to those in observational studies (14,15,18,24) (7–10 ng/mL).

Rejection rates and tacrolimus TC

Using lower tacrolimus TC could result in more ACR, especially early after transplantation, when there is increased graft immunogenicity. Nevertheless ACR after LT is often mild without significant detriment to graft and patient survival (7,30). Furthermore lower tacrolimus concentrations could translate into more immune activation, which is beneficial to induce tolerance in the long term (31). Only moderate/severe ACR is most likely to require specific treatment to prevent chronic rejection and graft loss.

Data from interventional studies: overall analysis:  Figure 3 shows rates of ACR in RCTs, which vary depending on how ACR is defined: most studies use histological confirmation after clinical suspicion, with ACR 20%–30%, while if protocol biopsies were used this was 20–50%. Rejection rates are unchanged over the last decade. There were 20 RCTs comprising 39 comparison arms and 5000 patients, defining ACR as histologically proven after clinical suspicion. In 21 arms the mean tacrolimus TC during the first month was 10–15 ng/mL, while 18 arms reported 6–10 ng/mL. In 5 RCTs (9 comparison arms; n = 719) which used protocol biopsies, 7 arms had tacrolimus TC <10 ng/mL during the first month after LT, and 2 arms had higher levels. Figure 4 shows no correlation between the mean TC and the occurrence of ACR either defined with protocol biopsies (r =−0.37; p = 0.32) or not (r = 0.11; p = 0.49). As this could be due to using other immunosuppressant drugs combined with tacrolimus, we further subdivided the analysis.

Figure 3.

Incidence of acute cellular rejection in liver transplant recipients among randomized clinical trials between 2002 and 2012 comparing protocols with tacrolimus (with standard or reduced dosage and in association or not with other drugs).

Figure 4.

Relationship between mean trough tacrolimus concentrations during the first month after liver transplantation and the occurrence of acute cellular rejection (defined by using protocol biopsies or histologically confirmed after clinical suspicion). Each data point represents one arm from the included trials (2002–2012).

Interventional studies aiming at lower TC of tacrolimus without differences in other immunosuppression: “minimizing strategy”:  Hitherto only one study has compared identical immunosuppressive regimens using different tacrolimus TC (32). A standard regimen using tacrolimus (TC 10–15 ng/mL), mycophenolate and tapering steroids was compared with an identical regimen but targeting tacrolimus TC at 5–8 ng/mL. ACR rates were similar (17%) (biopsy proven after biochemical suspicion requiring treatment) by 26 weeks after transplantation. As the primary objective was to evaluate pharmacokinetics, only 30 patients/arms were included, so a type 2 error is possible. However, results suggest that lower tacrolimus TC may be adequate to prevent ACR.

Other studies have also used lower tacrolimus TC without large differences in ACR. One (33) allowed a wider TC range (5–15 ng/mL) in both arms: ACR in patients receiving only tacrolimus and steroids was 26.6%. Another (29) compared tacrolimus monotherapy with tacrolimus and steroids using a target of 6–8 ng/mL in both arms: moderate/severe ACR using protocol biopsies at day 7 was 50% and 48.7%, respectively, rates within the range of other studies using protocol biopsies. However approximately half of patients had TC <6 ng/mL at the time of biopsy. Another study from the same group also using protocol biopsies and aiming for identical tacrolimus TC had ACR rates between 0% and 50% (only 20%–25% of patients had <6 ng/mL) (34). Importantly both studies (29,34) showed that the higher ACR rates did not affect patient or graft survival, and lower rates of renal dysfunction (creatinine ≥1.5 mg/dL) were seen (5% to 10%). Thus tacrolimus used in monotherapy or combined with steroids (without other interventional drugs) could be safe for long-term outcomes.

In HCV transplanted patients, tacrolimus monotherapy was compared with tacrolimus, azathioprine and steroids aiming at lower tacrolimus TC in both arms (5–10 ng/mL), and ACR was diagnosed by protocol biopsies (35,36). There were higher rates of moderate–severe ACR within the tacrolimus-based triple therapy group (48% vs. 29%) (35,36). Again these rates are within the range seen in studies using protocol biopsies, and the higher rate seen in the triple therapy arm was likely to be due to lower tacrolimus TC. Graft and patient survival were similar (36).

Interventional studies aiming at lower tacrolimus TC by adding another immunosuppressive drug: “renal sparing strategy”:  Five RCT (13,28,32,37,38) and one nonrandomized trial (39) compared protocols in which conventional and reduced TC of tacrolimus were in opposing arms (see Table 1), evaluating anti-IL2 or antithymocyte globulin, or adding mycophenolate. ACR was defined solely by histological confirmation after biochemical suspicion of rejection, which is an important limitation as liver function tests are not specific (40), and are only late markers (41) of histological ACR. Furthermore, the “clinical” definition of suspected rejection is not mentioned, or is different, between studies. With these caveats the meta-analysis of the five RCT including 917 patients (13,28,32,37,38) (quality assessment in Table 2) showed no significant differences for biopsy-proven ACR (RR = 0.92 [95% CI = 0.65–1.31]), between reduced (6–10 ng/mL) and conventional (>10 ng/mL) tacrolimus TC (Figure 5). The I2 test for heterogeneity was 54% (p = 0.07). The funnel plot did not reveal publication bias (Supporting Figure S1). Results did not substantially change when we repeated the analysis including only studies with optimal methodological quality (28,37) (RR = 0.82 [95% CI = 0.66–1.01]; I2= 59%, p = 0.12).

Table 2.  Quality assessment of randomized controlled trials comparing conventional (>10 ng/mL) and low (<10 ng/mL) trough concentrations of tacrolimus in opposite arms within the first month after liver transplantation
AuthorSequence generationAllocation concealmentBlindingIncomplete outcome dataSelective outcome reportingOther sources of bias
  1. *Measured outcomes were objectively assessed (serum creatinine for renal impairment and Banff grading for histological acute rejection); in the authors’ opinion these are unlikely to be influenced by the lack of blinding and do not influence the result of the meta-analysis.

Boudjema (28)YesYesYes*YesYesYes
 “Randomization was based on computer-generated lists”“Randomization lists were kept by the pharmacist of the coordinating center”.Not blinded but outcome is unlikely to be affectedMissing data balanced across groups  
Benítez(13)YesYesYes*YesYesNo
 “The randomization sequence was computed-generated”“The randomization sequence was kept in opaque sealed envelopes”Not blinded but outcome is unlikely to be affectedCauses of protocol discontinuation were not related with primary outcome (rejection) The study was prematurely stopped due to imbalance in outcome
Nashan (32)UnclearUnclearYes*NoYesYes
 The authors did randomization but the method used is not describedInformation about allocation concealment is not providedNot blinded but outcome is unlikely to be affected“Other reasons for withdrawal in the tacrolimus standard-dose group were insufficient therapeutic response (2 patients)”. This could be related to outcome  
Neuberger (37)YesYesYes*YesYesYes
 “using a randomization list with blocks of six patients”“Patients were randomized centrally”Not blinded but outcome is unlikely to be affected36% of patients discontinued the study prematurely but the reasons were unlikely related to outcome  
Yoshida (38)UnclearUnclearYes*YesYesYes
 The authors did randomization but the method used is not describedInformation about allocation concealment is not providedNot blinded but outcome is unlikely to be affectedCase analysis available  
Figure 5.

Forest plot diagram illustrating the risk of acute cellular rejection among five randomized controlled trials (n = 957) using reduced (<10 ng/mL) and conventional (>10 ng/mL) tacrolimus trough concentrations in opposite arms.

Overall among the six trials (these five RCT and the not randomized study), four (37–39) showed no statistical differences in ACR: 11%–23.2% in standard, and 14.9%–27% in reduced dose arms. In another RCT standard tacrolimus (TC 10–15 ng/mL during the first 3 months and 7–12 ng/mL thereafter), was compared to induction with antithymocyte globulin and reduced tacrolimus (TC 5–12 ng/mL during the first 3 months and slow weaning thereafter to achieve very low doses at 12 months) (13). As there was a significantly higher rejection rate in the antithymocyte induction group (66.7% vs. 31.2%; p = 0.03) the trial was prematurely stopped. Interestingly all patients developing rejection during weaning had TC <5 ng/mL suggesting that complete weaning is not possible, and TC under this threshold within 12 months are not safe. In the sixth study, 100 patients with “full dose tacrolimus” (TC >12 ng/mL for the first 6 weeks) were compared with 95 patients given mycophenolate and “reduced dose tacrolimus” (TC <10 ng/mL for the first 6 weeks). Both groups also received tapering steroids. ACR was decreased with lower tacrolimus TC (30% vs. 46%; p = 0.024) (28).

Data from observational studies:  The data here are heterogeneous. In a retrospective study comprising 248 patients who received a liver, kidney, lung or combined transplant the relationship between tacrolimus TC and ACR was significant but weak (14). In a prospective analysis of 111 LT patients tacrolimus TC (over a 7-day period before diagnosis of histological rejection) independently predicted ACR (RR = 0.86; p = 0.049), but as the AUROC was <0.5 no optimal predictive threshold could be derived (15). In contrast, a systematic review of three RCTs in LT (n = 721) (42–44) did not find any relationship between tacrolimus TC and ACR (17). Nevertheless, these results are barely interpretable because of different definitions of ACR between studies and the use of plasma levels of tacrolimus instead of whole-blood levels. Furthermore the first and only study which evaluated protocol biopsies in 90 patients (18), documented moderate–severe ACR in 41.1% at day 7, without correlation between tacrolimus TC and histological ACR (mean TC was 7.3 ± 4.4 with no-mild rejection and 7.5 ± 3.7 with moderate–severe rejection).

There are several important limitations in interpreting these results. Firstly, both the populations and immunosuppression protocols were heterogeneous; secondly most studies did not use protocol biopsies; thirdly the threshold for liver function test abnormalities defining a biochemical suspicion of ACR, varied between studies and even between centers within the same study (1). Lastly using a 7-day period (15) between measuring TC and correlating with histological confirmation of ACR, does not mirror clinical practice, and has little applicability.

Renal dysfunction related to tacrolimus after liver transplantation

Chronic renal failure occurs in 18% of liver transplant patients by 5 years: only intestinal transplantation is worse (9). In contrast to the weak relationship between tacrolimus TC and ACR, the relationship with acute and chronic renal dysfunction is well established. Indeed early exposure to tacrolimus after LT is particularly deleterious, affecting long-term outcomes (9,10).

Data from interventional studies: overall analysis:  Among RCTs, renal dysfunction is documented in 20–50% (Figure 6), although three studies reported over 60% (36,39,45), two of them (39,45) had very high exposure to tacrolimus (>12 ng/mL) during the first month after LT. As shown in Figure 6 there is an increasing trend of higher rates of renal dysfunction at 1 year after LT.

Figure 6.

Incidence of renal impairment in liver transplant recipients among randomized clinical trials between 2002 and 2012 comparing protocols with tacrolimus (with standard or reduced dosage and in association or not with other drugs).

There were 16 RCTs including 31 comparison arms and 4114 patients, documenting renal impairment between 3 and 6 months (17 arms), and at 1 year (14 arms). In 18 comparison arms the mean tacrolimus TC recorded (during the first month after LT) was within current recommended levels (10–15 ng/mL; mean 11.7 ± 1.1 ng/mL). In the other 13 comparison arms tacrolimus TC were lower (6–9.9 ng/mL; mean 8.96 ± 0.9 ng/mL). Rates of renal impairment in RCTs positively correlated with mean tacrolimus TC during the first month after LT (r = 0.58; p = 0.001), especially for studies reporting renal dysfunction at 1 year (r = 0.73; p = 0.003) (Figure 7).

Figure 7.

Relationship between mean trough tacrolimus concentrations during the first month after liver transplantation and renal impairment in randomized controlled trials (2002–2012). Each data point represents one study arm from the included trials.

Interventional studies aiming at lower TC of tacrolimus without differences in other immunosuppression:minimizing strategy”:  The single study using identical protocols evaluating 60 patients either with standard tacrolimus TC (aiming 10–15 ng/mL), or reduced dose (TC 5–8 ng/mL), found that creatinine clearance was higher with reduced TC, but this difference was already at the baseline so this result is uninterpretable (32). Interestingly two further RCTs from the same group (29,34) reported renal impairment (serum creatinine ≥1.5 ng/mL) under 11% when aiming for lower tacrolimus TC (6–8 ng/mL).

Interventional studies aiming at lower tacrolimus TC by adding another immunosuppressive drug: “renal sparing strategy”:  Two RCT (28, 37) (n = 712) compared reduced and conventional tacrolimus TC within the first month after LT and reported rates of renal impairment, both at 1 year (quality assessment in Table 2). In the metaanalysis (Figure 8), lower tacrolimus TC (6–10 ng/mL) led to less renal impairment (RR = 0.51 [95% CI = 0.38–0.69]) compared with conventional tacrolimus TC (>10 ng/mL). I2 was 0% (p = 0.44). The funnel plot did not reveal publication bias (Supporting Figure S1).

Figure 8.

Forest plot diagram illustrating the risk of renal impairment at 1 year among two randomized controlled trials (n = 712) using reduced (<10 ng/mL) and conventional (>10 ng/mL) tacrolimus trough concentrations in opposite arms.

In other RCT renal function was compared between standard immunosuppression (n = 76) comprising tacrolimus (TC 10–15 ng/mL during the first month), mycophenolate and steroids, and an experimental arm (n = 72) receiving the same protocol but adding daclizumab and delaying use of tacrolimus until days 4–6 after LT (TC 4–8 ng/mL) (38). By day 3 after LT, the glomerular filtration rate had decreased 25 mL/min from baseline in the standard arm, while the experimental arm (who had not received tacrolimus as yet) had an improved glomerular filtration rate by 12 mL/min (p = 0.001), a difference that remained statistically significant at 1 month after LT.

With a similar design without mycophenolate, and substituting daclizumab for basiliximab in 45 patients, a not randomized trial showed renal impairment rates of 67% at 3 months in the standard arm, and 26% in the delayed reduced arm (TC 5–10 ng/mL), (p < 0.001)(39). In 195 patients aiming for tacrolimus TC <10 ng/mL early after LT, a beneficial effect on renal function was still present at 1 year after (42% to 24%; p = 0.009) (28) confirming the importance of the early period after LT with respect to renal toxicity. Figure 6 shows that another RCT which used a reduced tacrolimus TC within the experimental arms (comprising daclizumab, tacrolimus, mycophenolate and steroids in one, and tacrolimus, mycophenolate and steroids in the other) (37), found a nonstatistical trend for less renal impairment in this group, a result likely due to a similar tacrolimus TC in both arms (differences <1 ng/mL) (37). Thus there was no appropriate evaluation of lower tacrolimus TC, and the effect of the additional immunosuppressive agent made the regimen more immunopotent and potentially increased adverse effects. Similar results were obtained in a RCT allowing reduced tacrolimus TC but only after the first month after LT (33).

Data from observational studies:  In 111 patients with a follow-up of 12 weeks, 43.2% experienced adverse effects due to tacrolimus. Multivariately tacrolimus TC were independently related to nephrotoxicity (OR 1.27; p < 0.001), and when higher than 11.6 ng/mL (common in the early postransplant period using standard target TC) the positive predictive value was 64% (15).

A retrospective study of 96 patients evaluated the relationship between nonalcoholic steatohepatitis (NASH) as the indication for LT and subsequent renal dysfunction (46). Acute kidney injury within the first month was higher among NASH patients (50% vs. 29%), as was persistent kidney injury after 3 months (31.2% vs. 8.3%). Interestingly despite more use of mycophenolate among the NASH group (40% vs. 21%), tacrolimus TCs were similar between the groups at 3 months (8.4 ng/mL vs. 7.7 ng/mL; p = 0.14) and 2 years (8.0 ng/mL vs. 8.1 ng/mL; p = 0.95) after LT. Thus, NASH patients may be particularly susceptible to renal dysfunction and should be considered for tacrolimus TC reduction. Again adding another drug (mycophenolate) did not result in lower tacrolimus TC, so that a real “test” of renal sparing was not achieved. In addition, the early exposure to tacrolimus was not recorded and differences in this aspect were not evaluated regarding renal outcomes.

Outcomes were reported in 124 patients receiving tacrolimus monotherapy (47) (30 from a RCT (48) and 94 from a consecutive prospective cohort), with a median follow-up of 8 years. Mean tacrolimus TC were between 7 ng/mL and 10.3 ng/mL during the first 3 weeks but 34% had peak tacrolimus TC >15 ng/mL, and 45% required dose reduction. Renal dysfunction (glomerular filtration rate <60 mL/min) occurred in 36% of patients. However long-term renal support was only needed in one patient (0.8%), a rate substantially lower than that reported by Ojo et al. (9) (18.1%) for patients transplanted during the same period with similar follow-up.

Discussion

In LT there is a different clinical impact of rejection compared to other organs (4,5), but recommended tacrolimus TC remain identical to kidney transplantation, in which ACR often carries higher risk of irreversible graft damage. An immediate consequence is that tacrolimus-related complications are frequent, which is particularly important given the correlation between early exposure to tacrolimus and renal impairment (Figure 7). Our systematic review of RCTs with 1-year follow-up showed an even stronger correlation. This confirms a premise for this review that tacrolimus early exposure is critical in determining long-term renal dysfunction. Indeed renal impairment occurs in up to 60% of patients (28,39,49,50) and is independently related to mortality (9). Preventing renal impairment is vital, as once established, reducing tacrolimus dosage (even by 50%) or conversion to non nephrotoxic drugs such as sirolimus or mycophenolate, only stabilizes renal dysfunction but rarely reverses it (51–55). The meta-analysis showed that lower tacrolimus TC after LT (6–10 ng/mL) led to a twofold reduction in renal impairment. Thus lower tacrolimus TC could counteract the recent trend of increasing rates of renal dysfunction at 1 year after transplantation (Figure 5).

Our review shows that lower tacrolimus TC do not adversely affect ACR. Several RCTs comparing standard and lower tacrolimus TC show no difference in terms of rejection (32,37–39), or even an advantage for the lower TC arm (28). Importantly in the only study directly comparing two identical protocols but varying tacrolimus TC, there was no difference in ACR (32). In our evaluation of RCTs there was no significant correlation between the mean tacrolimus TC and ACR (Figure 4). The meta-analysis showed no differences in ACR when targeting tacrolimus TC <10 ng/mL or >10 ng/mL within the first month after LT. There was a trend toward significant heterogeneity (I2= 54%; p = 0.07) which may be explained by the lack of consensus in the definition of clinical suspicion of rejection and also because one study aimed to use very low tacrolimus TC (<6 ng/mL) in the “reduced dose” arm (13). The complete weaning of tacrolimus, or achieving very low concentrations during the first weeks after LT carries a higher risk of rejection (12,13), which could be deleterious.

There are other potential benefits of using lower tacrolimus TC early after transplantation. Several animal models, and one prospective study in humans documented a positive correlation between tacrolimus exposure and hepatocellular carcinoma recurrence (56–58). Secondly HCV patients may develop less fibrosis as shown in a RCT using lower tacrolimus TC (5–10 ng/mL) (36), compared to other studies (59,60). There may be a benefit in not suppressing ACR completely in terms of graft tolerance, since rejection and tolerance are along a continuum (31,61).

We searched to find evidence for the optimal tacrolimus TC after LT. This proved to be difficult: most RCTs did not use equipotent regimes, as they were designed to test new agents, and not specifically lower doses of tacrolimus. Whether results would be similar without additional agents cannot be definitely ascertained, but it is likely from the evidence taken as a whole. However, in observational studies, there was great variability in the relationship between tacrolimus TC and the risk of ACR. This may be due to the definition of ACR used, which was based on clinical suspicion in most studies. Protocol biopsies would provide a more objective assessment but these are seldom performed.

However this systematic review suggests that to reduce renal impairment and not to increase moderate–severe ACR simultaneously, it would be preferable to target tacrolimus TC between 6 and 10 ng/mL during the first 4–6 weeks after LT. Then a progressive reduction of dosage should be accomplished to achieve TC between 4 ng/mL and 8 ng/mL in the long term. Studies with large cohorts of liver transplant patients with protocol biopsies are needed to determine the lowest threshold of TC possible, which may vary for different etiologies (24,46) and immunosuppression protocols. Future RCTs using tacrolimus should target lower TC, which will require regulatory changes.

Acknowledgments

M.R.P. is a recipient of a grant from the Institute Maimónides for Biomedical research of Córdoba (IMIBIC).

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary