Interleukin 2 receptor antagonists for liver transplant recipients: A systematic review and meta-analysis of controlled studies

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Interleukin 2 receptor antagonists (IL-2Ra) are frequently used as induction therapy in liver transplant recipients to decrease the risk of acute rejection while allowing the reduction of concomitant immunosuppression. We conducted a systematic review of prospective, controlled studies to test the hypothesis that the use of IL-2Ra is associated with a decrease in acute rejection and/or a decrease in the side effects of concomitant medication. We performed a search of all major databases and secondary sources from inception to December 2010. Random effects models were used to assess the incidence of acute rejection, graft loss, patient death, and adverse side effects, with or without IL-2Ra. Subgroup analysis and meta-regression were used to explore differences in effect and sources of heterogeneity. Eighteen studies (13 randomized and 5 nonrandomized) met the inclusion and exclusion criteria. Acute rejection at 12 months or later favored the use of IL-2Ra (relative risk [RR] 0.83; 95% confidence interval [CI] 0.76-0.94) and steroid-resistant rejection was also less frequent in patients receiving IL-2Ra (RR 0.66; CI 0.48-0.91). Graft loss and patient death did not differ significantly between treatments. Patients who received IL-2Ra in addition to reduced or delayed calcineurin inhibitors had better renal function (mean difference of estimated glomerular filtration rate: 6.29 mL/min; CI 1.66-10.91) and a lower incidence of renal dysfunction (RR 0.46; CI 0.27-0.78). The use of IL-2Ra was also associated with a lower incidence of posttransplant diabetes mellitus, whereas the incidence of other adverse events was similar. Conclusion: The use of IL-2Ra is associated with a lower incidence of acute rejection after transplantation. Concomitant immunosuppression can be reduced, avoiding long-term side effects of immunosuppression. (HEPATOLOGY 2011;).

The advent of new immunosuppressive agents such as rapamycine and monoclonal antibodies in the 1990s raised hopes of further improving the long-term outcome of transplant patients. Only recently, the effects of rapamycine were systematically reviewed in Hepatology, where it was reported that rejection rates and renal function were not significantly different with or without rapamycine.1

Monoclonal antibodies targeting the interleukin 2 (IL-2) receptor (Il-2R) are now used in every fourth liver transplant recipient in the USA2 and are also frequently used in Europe. Two IL-2R antibodies (IL-2Ra), daclizumab and basiliximab, were commercially available, but daclizumab was recently withdrawn from the market for commercial reasons. Daclizumab, but not basiliximab, had been approved for use in liver transplant recipients, although the latter is still used for this purpose in many transplant centers, especially following the withdrawal of daclizumab.

IL-2Ra specifically bind and block the IL-2R α-chain (which corresponds to CD25), which is present only on the surface of activated T-lymphocytes.3 The IL-2 signal is essential for the activation of lymphocytes; it induces second messenger signals to stimulate T cells to enter the cell cycle and proliferate, resulting in clonal expansion and differentiation. The commercially available IL-2Ra are both monoclonal anti-CD25 immunoglobulin G (IgG) antibodies, but their structure and synthesis are different. Daclizumab is a humanized antibody built by total gene synthesis using oligonucleotides,4 whereas basiliximab is a chimeric murine-human antibody.5 The competitive block of IL-2R, and thereby of IL-2-mediated activation, lasts for 4 to 12 weeks, depending on the antibody and the administration protocol.3

Currently, IL-2Ra are mainly used as induction agents, to prevent acute rejection and minimize early graft injury, but they have also been used to treat acute rejection. In renal transplantation, large randomized trials have shown that both IL-2Ra reduce the incidence of acute rejection and have a relatively good toxicity and safety profile.5, 6 But there have also been some concerns about the long-term effects, especially regarding posttransplant lymphoproliferative disorders (PTLD) and other malignancies.3 The effects of IL-2Ra have also been evaluated in a meta-analysis of kidney transplant recipients.7 The results showed that induction with IL-2Ra significantly reduces the risk of acute rejection but has no effect on graft or patient survival.

A first nonsystematic review of the literature showed that in liver transplant patients, IL-2Ra are not only used in addition to standard immunosuppression but are mainly used to reduce other immunosuppressive drugs, such as calcineurin inhibitors (CNI) and corticosteroids, thereby possibly decreasing the incidence and severity of their adverse effects. We have therefore structured this meta-analysis into three separate comparisons as follows: (1) comparison of IL-2Ra versus placebo or no treatment; (2) comparison of IL-2Ra with reduced and/or delayed CNI versus placebo or no IL-2Ra treatment in combination with standard immunosuppression; and (3) comparison of IL-2Ra and reduced or no corticosteroids versus placebo or no IL-2Ra treatment in combination with standard immunosuppression.

Abbreviations

ACA, available-case-analysis; AE, adverse event; CMV, cytomegalovirus; CNI, calcineurin inhibitor; eGFR, estimated glomerular filtration rate; GFR, glomerular filtration rate; HCV, hepatitis C virus; IL-2R, interleukin-2 receptor; IL-2Ra, interleukin-2 receptor antagonists; ITT, intention-to-treat analysis; LOCF, last-observation-carried-forward; MD, mean difference; MDRD, modification of diet in renal disease; MMF, mycophenolate mofetil; NNT, number needed to treat; PTDM, post-transplant diabetes mellitus; PTLD, post-transplant lymphoproliferative disease; REML, restricted maximum likelihood; SAE, serious adverse event.

Materials and Methods

The methods of literature search, the inclusion and exclusion criteria, outcome measures, and methods of statistical analysis were defined in a protocol according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions.8 We also used the Preferred Items for Systematic Reviews and Meta-Analysis (PRISMA) and Meta-Analysis of Observational Studies in Epidemiology (MOOSE) recommendations for study reporting.9, 10

Literature Search

A systematic literature search was performed without language restrictions from inception to December 2010 in the following databases: Medline/PubMed, Embase, Transplant Library, and Cochrane Library. The keywords used were “liver transplantation,” “interleukin 2 receptor inhibitor/antagonist,” “basiliximab,” “daclizumab,” “simulect,” “zenapax,” and abbreviations thereof, combined with appropriate Boolean operators. The reference lists in all identified trials were examined for further relevant articles. To locate unpublished trials, we contacted authors of included reports and drug manufacturers where appropriate. Non-English articles were translated by a medical specialist fluent in the respective languages.

Inclusion and Exclusion Criteria

All prospective, controlled, experimental (randomized), and observational (nonrandomized) studies in which IL-2Ra induction therapy in liver transplant recipients was compared with placebo or no treatment were included. For comparison 1, we included only studies in which IL-2Ra was compared to placebo or no treatment with otherwise the same immunosuppressive treatment in both study arms. For comparison 2, we included studies with reduced and/or delayed CNI in combination with IL-2Ra; and in comparison 3, we included studies with reduced corticosteroids in combination with IL-2Ra. Other immunosuppressive medication, e.g., mycophenolate mofetil, had to be the same in both treatment arms.

Studies with historical controls were also included, but we excluded studies in which both cohorts were assessed retrospectively. We also excluded noncontrolled studies and pharmacological studies that did not provide data on clinical outcome measures because of their very short follow-up time. With regard to patient selection, we excluded trials with patients undergoing multiorgan transplantation or retransplantation.

Outcome Measures

The primary outcomes analyzed were graft loss, acute rejection, steroid-resistant rejection, and death. Other outcome measures assessed were renal dysfunction (serum creatinine and/or estimated glomerular filtration rate [eGFR]), de novo malignancy (excluding recurrence of hepatocellular carcinoma), PTLD, infectious complications, including cytomegalovirus (CMV) infection, new onset of metabolic and cardiovascular disorders, such as hypertension, hyperlipoproteinemia, and posttransplant diabetes mellitus (PTDM), and all other adverse reactions (as a direct consequence of drug treatment).

Contribution of Reviewers

There were four reviewers (A.D.G., A.O., N.H., N.B.). The literature search strategy was designed and performed by three reviewers (A.D.G., A.O., N.H.). The search results were combined in an open source reference management software (JabRef v. 2.6.0). Publications were screened independently by three reviewers (A.D.G., N.H., N.B.). Disagreement and any discrepancies were resolved by discussion (A.O. with A.D.G., N.H., N.B.). Data extraction was performed by two reviewers (A.D.G., N.H.), using a standardized form. A training set was used to validate data extraction.

Study Quality

Quality of studies was assessed independently by two reviewers (A.D.G., N.H.) without blinding to journal and authorship. The quality items assessed were blinding, randomization, allocation concealment, intention-to-treat analysis (ITT), completeness of follow-up, and the method of handling missing values. Assessment was performed according to definitions stated in the Cochrane Handbook.8 Furthermore, completeness of follow-up was defined as the number of patients that were not lost to follow-up. We either reported completeness of follow-up as stated by the authors or tried to extract approximate data from the text. Methods of handling missing values are stated as reported by the authors.

Data Extraction

All available data for the described outcome measures were extracted at all available timepoints from individual trials. When data were not explicitly stated in the text but given in graphical form, we used calipers to extract data from the appropriate graphs. Data of continuous variables given only in median and/or interquartile range (IQR) were converted to mean and standard deviation according to methods stated in the Cochrane Handbook.8 Data given only in median, minimum, and maximum were excluded from the analysis.

In contrast to kidney transplants, it has been shown that morphological signs of rejection in protocol biopsies of transplanted livers without clinical correlates require no treatment and have no long-term adverse effects.11 Therefore, we only included treated acute rejections in the primary analysis, when the reported acute rejection was stratified into “treated” and “nontreated.” When data on outcome measures were not provided, the authors were contacted to provide more data.

Data Analysis

We expressed the results of dichotomous outcomes as relative risk (RR) with values of <1 favoring IL-2Ra, and continuous outcomes as weighted mean differences (MD), both with 95% confidence intervals (CI). We performed the analysis with both random and fixed effects and found no relevant differences. Results reported here used the random effects model, as this is more conservative in the presence of heterogeneity.12

For the random effects models the amount of residual heterogeneity (i.e., τ2) was estimated by the restricted maximum likelihood (REML) method.13 Confidence intervals for τ2 were obtained by the Q-profile method.14 The model parameters were estimated by way of weighted least squares, with weights equal to the inverse sum of the variance of the estimate and the estimate of the residual heterogeneity. Then Wald-type tests and confidence intervals were obtained for the parameter estimates.13

We analyzed heterogeneity among studies using Cochrane's Q test and calculating I2 to measure the proportion of total variation due to heterogeneity beyond chance.15 When we observed heterogeneity, we also performed regression diagnostics of random effects models by computing and inspecting the externally Studentized residuals, Cook's distance, and the weights during the model fitting to identify outlying and/or influential studies.13 Residual heterogeneity was further explored by estimating τ2 and the test statistic Q when each study was removed in turn (leave-one-out deletion).13

We performed subgroup analyses and meta-regression for all primary outcomes and when significant heterogeneity was observed. Subgroups and factors (for meta-regression) defined a priori were methodological quality of trial (i.e., randomized versus nonrandomized), comparison group, type of IL-2Ra, type of CNI, and use of mycophenolate mofetil (MMF). For the primary analysis we pooled effect measurements from trials with different follow-up time; but timepoint of measurement (grouped by 3 to 6 months versus 12 months and later) was evaluated in subgroup analysis and meta-regression to explore possible effects of time.

Effects of study-level covariates on overall rejection rate were assessed by fitting generalized linear mixed models (GLMM).16

Publication bias was assessed by funnel plots,17 the trim-and-fill method,18 and tests for funnel plot asymmetry.13 When publication bias was suspected we fitted random effects models with data augmented by the trim-and-fill method.13 The R environment for statistical computing (v. 2.11.0)19 with packages “metafor” (v. 1.4-0)13 and “lme4” (v. 0.999375-37)16 were used for all analyses.

Results

Literature Search

Database searches and other resources (mainly conference proceedings) yielded 1,233 entries (see Fig. 1), of which 261 were excluded as duplicates. Of the 972 publications that qualified for abstract review, 852 were excluded primarily because they were not controlled trials, IL-2Ra were not compared in the study, or they were not conducted in patients undergoing first liver transplantation. The remaining 120 publications underwent full article review (where available) and a further 86 publications were excluded mainly because they did not compare IL-2Ra, they were entirely retrospective, or they were performed in pediatric patients. Three trials20-22 were excluded because information, e.g., regarding the methodological quality, was inconclusive and we could not obtain further information by contacting the authors.

Figure 1.

Flow chart of publication search and selection. IL-2Ra, interleukin-2 receptor antagonist; IS, immunosuppression; OLT, orthotopic liver transplantation. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.].

A total of 18 studies qualified for inclusion in this review23-40 with a total of 2,961 randomized patients. For three trials25, 37, 39 only an abstract was available, whereas for the remaining 15 trials23-24, 26-36, 40 a full text publication was obtained with 16 additional reports (e.g., conference abstracts, follow-up reports). In case of multiple reports on the same study we cited the first full-text publication as the index publication. Eleven authors of reports with missing information were contacted but either did not reply or could not provide further information. One publication was only available in Chinese32; all other were reports were in English. The proportion of interrater agreement for study selection was 98.7% with a kappa index41 of 0.83.

Included Studies

Table 1 shows the characteristics of the included studies. Five trials25, 28, 32, 35-36 compared IL-2Ra to placebo or no treatment without modification of concomitant immunosuppressive medication (comparison 1). Six trials24, 26, 31, 34, 39-40 compared IL-2Ra in combination with reduced and/or delayed CNI to placebo or no treatment with standard immunosuppression (comparison 2). Seven trials23, 27, 29-30, 33, 37-38 compared IL-2Ra in combination with early discontinuation or reduction of steroids to placebo or no treatment with standard immunosuppression (comparison 3). Four studies were restricted to patients with chronic hepatitis C virus (HCV) infection25, 29-30, 37 and in five studies patients with moderate to severe renal dysfunction were excluded.24, 33-35, 40

Table 1. Characteristics of Included Trials Stratified by the Three Pre-specified Comparison Groups
Trial (Author & Year) [citation]Patient subgroupsSample SizeAge*Sex (male)Type of IL-2RACNIMMFLength of Follow-up
ExpContExpContExpCont
  • Abbreviations: Exp, experimental group; Cont, control group; HCV, hepatitis C virus; Bas, basiliximab; Dac, daclizumab; CNI, calcineurininhibitor; Cya, cyclosporine A; Tac, tacrolimus; MMF, mycophenolate mofetil; ns, not stated.

  • *

    Age is given in mean ± standard deviation if available

  • Time given in months

  • Only abstract available

  • §

    Trial with delayed and reduced CNI in experimental arm versus delayed CNI in control group

  • #

    Trial with two separate cohorts: cohort 1 did not receive MMF, cohort 2 received MMF

  • •Only mean follow-up available

  • Mean follow-up in experimental group 20 months and in control group 7 months

Comparison 1: IL-2Ra vs placebo/no treatment
Fasola 200525HCV-positive4624nsnsnsnsDacTacyes12
Innocenti 200328 24105448nsnsDacCyayes20/7
Lu 200632 4027nsnsnsnsDacTacyes6
Neuhaus 200235normal renal function1881934950.2117124BasCyano12
Schmeding 200736 514849.449.62727BasTacno> 36
Comparison 2: IL-2Ra and delayed and/or reduced CNI vs placebo/no treatment and standard immunosuppressive co-medication
Calmus 201024normal renal function9810152.3 ± 9.653.6 ± 9.27774DacTacyes24
Heffron 200126 544751.5 ± 6.344.1 ± 9.4nsnsDacTacyes12
Lin 200531 271849.345.52214BasTacyes6
Neuberger 2009§34normal renal fucntion1681685554116109DacTacyes12
Yan 200439 2424nsnsnsnsBasCyayes3
Yoshida 200540normal renal function727653 ± 8.7852.4 ± 8.045050DacTacyes12
Comparison 3: IL-2Ra and minimized steroids or no steroids vs placebo/no treatment and standard immunosuppressive co-medication
Boillot 200523 35134750.9 ± 10.451 ± 9.8239238DacTacno3
de Simone 200737HCV-positive9595nsnsnsnsBasCyayes12
Humar 200727 838352.351.8nsnsBasTacyes16
Kato 2007 cohort 1#29HCV-positive151653.5 ± 8.150.6 ± 6.8811DacTacno12
Kato 2007 cohort 2#29HCV-positive162351.3 ± 650.0 ± 5.81318DacTacyes12
Klintmalm 200730HCV-positive1537951.5 ± 7.451.4 ± 7.810960DacTacyes12
Lupo 200833normal renal function262150.553.72318BasCyano22
Washburn 200138 151548.1 ± 8.656.4 ± 8.5118DacTacyes18

In most studies daclizumab was used for induction.23-26, 28-30, 32, 34, 38, 40 Concomitant immunosuppressive medication (see Supporting Table 1 for details) included MMF in most studies,24-32, 34, 37-40 prednisolone in all studies, and calcineurin inhibitors, i.e., tacrolimus in 13 studies23-27, 29-32, 34, 36, 38, 40 and cyclosporine A in the remaining five studies.28, 33, 35, 37, 39 One study was divided into two cohorts because of different concomitant immunosuppression.29 Most trials had a follow-up of 12 months or longer, but four trials had a study duration of only 3-6 months.23, 31-32, 39

Quality of Included Studies

Table 2 shows the quality assessment of the included studies. The risk of bias is summarized in Supporting Fig. 1. Two studies35, 39 were randomized, double-blinded, and placebo-controlled. Of the remaining 16 studies, 11 were randomized,23-25, 29-30, 33-34, 36-38, 40 three were nonrandomized,27, 28, 31 and whether randomization was performed or not could not be determined in two studies26, 32 (for the analysis, these studies were assumed to be nonrandomized). All studies were entirely prospective except for one trial27 in which a prospective experimental group was compared to historical controls. Of the randomized trials, allocation concealment was found to be adequate in four trials.24, 33-35

Table 2. Summary of Quality Assessment of Included Trials
Trial (Author&Year)BlindingRandomizedAllocation ConcealmentITT analysisMissing ValuesCompleteness of follow up§Control group
Exp. (%)Control (%)month
  • Abbreviations: ITT, intention-to-treat analysis; LOCF, last observation carried forward; ACA, available case analysis; na, not applicable; ns, not stated.

  • *

    authors reported that IIT was performed but also stated conditions that must be met for patient to be included in analysis, such as “patient received at least one dose of medication” and/or “at least one follow-up available”

  • ITT analysis was assumed because authors reported on all patients in at least one analysis at the end of study

  • some continuous follow-up data imputed by linear regression

  • §

    as stated by authors or calculated from available data

Comparison 1
Fasola 200525noyesunclearyesnsnsnsnsconcurrent
Innocenti 200328nononaunclearnsnsnsnsconcurrent
Lu 200632nsnsunclearunclearnsnsnsnsconcurrent
Neuhaus 200235yesyesadequateno*ns84.682.912concurrent
Schmeding 200736noyesunclearunclearnsnsnsnsconcurrent
Comparison 2
Calmus 201024noyesadequateno*ACA899512concurrent
Heffron 200126nsnsunclearunclearnsnsnsnsconcurrent
Lin 200531nononayesnsnsnsnsconcurrent
Neuberger 200934noyesadequateno*LOCF6868.212concurrent
Yan 200439yesyesunclearyesnsnsnsnsconcurrent
Yoshida 200540noyesunclearunclearACA77.884.212concurrent
Comparison 3
Boillot 200523noyesunclearno*ns8173.83concurrent
de Simone 200737noyesunclearunclearnsnsnsnsconcurrent
Humar 200727nononaunclearnsnsnsnshistorical
Kato 2007 cohort 129noyesunclearnoACA60nsnsconcurrent
Kato 2007 cohort 229noyesunclearnoACA64nsnsconcurrent
Klintmalm 200730noyesunclearyesLOCFnsnsnsconcurrent
Lupo 200833noyesadequateyesLOCFnsnsnsconcurrent
Washburn 200138noyesunclearyesnsnsnsnsconcurrent

In seven study reports26-28, 32, 36, 37, 40 the patient population for the statistical analysis was not clearly defined. ITT analysis was stated and performed in two studies,30, 33 and we assumed ITT analysis in three studies25, 31, 39 because the authors report on all patients at the end of the study. Four authors23, 24, 35, 34 defined patient subsets (e.g., “modified ITT” or “full analysis set”) for the primary analysis and, for example, excluded patients that did not receive any study medication or patients that did not have any follow-up. One author did not perform ITT analysis and did not state the reasons for exclusions of patients from the analysis.29

Most authors23, 25-28, 31-32, 35-39 did also not state how missing values were handled. Available case analysis for continuous variables was evident from three studies24, 40, 29 and imputation by last-observation-carried forward was reported in three other studies.30, 33, 34

Outcomes

Acute Rejection.

Reduction of acute rejection favored the use of IL-2Ra (RR 0.84; CI 0.76-0.94; P = 0.002; 19 trials/cohorts) and the effect is seen in randomized and nonrandomized studies (Fig. 2). Stratifying trials by time of measurement showed a significant reduction of acute rejection with IL-2Ra at 12 months or later (RR 0.83; CI 0.74-0.94; P = 0.004; 14 trials/cohorts) but not at 3-6 months (Supporting Fig. 2), although this study-level covariate was not significant in the meta-regression (P = 0.76, Table 3). Furthermore, subgroup analysis stratified by the type of IL-2Ra used (Supporting Fig. 3) shows significant effect of daclizumab (RR 0.82; CI 0.71-0.95; P = 0.007; 12 trials/cohorts) but not basiliximab (RR 0.87; CI 0.73-1.03; P = 0.11; seven trials), but this does not seem to be a systematic effect because meta-regression does not indicate a significant effect of type of IL-2Ra (P for test of moderators = 0.67; Table 3) and inspection of the funnel plot (Supporting Fig. 4) shows that both types of studies are distributed similarly.

Figure 2.

Forest plot of acute rejection rate stratified by randomization status of the studies. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.].

Table 3. Meta-regression, to Explore Possible Confounding Effects on Summary Risk of the Primary Outcomes
Potential confounderAcute RejectionPatient DeathGraft Loss
Ratio of relative risk [95% CI]P* valueτ2 [95% CI]P# valueRatio of relative risk [95% CI]P* valueτ2 [95% CI]P# valueRatio of relative risk [95% CI]P* valueτ2 [95% CI]P# value
  • Results are expressed as the ratio of relative risk with each potential modifier, compared with the relative risk of the reference category of that confounder. Ratios <1 correspond to a smaller relative risk for trials with the potential confounder, so reflecting larger benefit for that characteristic.

  • *

    P value for test of moderators;

  • Estimated residual heterogeneity;

  • #

    P value for test of heterogeneity.

Model without covariatesNANA0 [0 to 0.057]0.70NANA0 [0 to 0.21]0.70NANA0.005 [0 to 0.36]0.68
Daclizumab vs Basiliximab0.96 [0.77 to 1.19]0.670.006 [0 to 0.070]0.641.37 [0.9 to 2.08]0.130 [0 to 0.20]0.771.12 [0.66 to 1.88]0.650.01 [0 to 0.47]0.65
Tacrolimus vs Cyclosporine A1.05 [0.83 to 1.33]0.660 [0 to 0.067]0.641.33 [0.84 to 2.11]1.330.004 [0 to 0.18]0.730.98 [0.53 to 1.82]0.940.02 [0 to 0.47]0.60
MMF vs no MMF0.83 [0.69 to 1.01]0.060 [0 to 0.038]0.820.92 [0.57 to 1.48]0.710.004 [0 to 0.31]0.620.77 [0.48 to 1.24]0.260 [0 to 0.45]0.69
Comparison 2 vs Comparison 10.77 [0.59 to 1.01]0.09 0.090 [0 to 0.051] 0 [0 to 0.051]0.83 0.831.18 [0.6 to 2.32]0.79 0.790.035 [0 to 0.29] 0.035 [0 to 0.29]0.58 0.580.75 [0.32 to 1.76]0.71 0.710.02 [0 to 0.57] 0.02 [0 to 0.57]0.58 0.58
Comparison 3 vs Comparison 11.01 [0.82 to 1.25]0.98 [0.54 to 1.76]0.97 [0.49 to 1.92]
Measurement at 12 months/later vs 3 to 6 months0.96 [0.75 to 1.23]0.720 [0 to 0.069]0.640.98 [0.94 to 1.02]0.280 [0 to 0.25]0.710.60 [0.38 to 0.94]0.030 [0 to 0.22]0.86

Meta-regression also showed that concomitant use of MMF (in both arms) seems to amplify the effect of IL-2Ra (ratio of RR 0.83; CI 0.69-1.01; P = 0.06). Analysis of the type of CNI did not show significant effects, but trials in comparison 2 also had a lower RR compared to trials in comparison 1 (Table 3). This effect may be explained by the fact that MMF was used in all trials in comparison 2. After adjusting for MMF, the effect in comparison 2 alone is no longer seen.

We did not observe significant heterogeneity in any of the analyses and observed only marginal changes of residual heterogeneity in meta-regression (Table 3). However, we observed considerable heterogeneity of overall rejection rates (i.e., the sums in the experimental and control groups of each trial) ranging from over 55%36 to less than 10%38 (see Supporting Table 2). Similar differences have also been observed in other meta-analyses concerning organ transplantation.42 We sought sources of heterogeneity using GLMM and found that the overall rejection rate was significantly higher in studies that required protocol biopsies (OR 3.10; CI 1.93-4.99; P < 0.001; 19 trials/cohorts). These studies reported not only treated rejections but also histological signs of rejection without clinical correlates. Subgroup analysis of trials with and without protocol biopsies (five trials/cohorts with 435 patients and 14 trials with 2,526 patients, respectively) showed a significant effect on acute rejection only in studies without protocol biopsies (RR 0.81; CI 0.71-0.92) but not in studies that performed protocol biopsies (RR 0.92; CI 0.76-1.11). Meta-regression does not provide any evidence for a significant effect of protocol biopsies on acute rejection (P value for test of moderators 0.26) and the inspection of the funnel plot (Supporting Fig. 5) shows that the distribution of both groups of studies is comparable.

Furthermore, the analysis of overall rejection rate showed that the use of MMF decreased the incidence of acute rejection in both study arms (OR 0.49; CI 0.31-0.77; P = 0.002; 19 trials/cohorts). After adjustment for use of MMF and type of biopsy the effect of IL-2Ra was still highly significant (OR 0.76; CI 0.64-0.90; P = 0.002; 19 trials/cohorts).

Funnel plot analysis for acute rejection showed significant asymmetry (P = 0.01). Using the trim-and-fill method we augmented the data (Supporting Fig. 6) and these supposedly missing studies all had a risk ratio above 1. However, the random effects model with the augmented data still showed a significant effect for use of IL-2Ra (RR 0.89; CI 0.80-0.98; P = 0.02).

Steroid-Resistant Rejection.

Six trials reported data on steroid-resistant rejection23-24, 27, 35-36, 39 measured at 3,23, 39 6,24 or 12 and more months.27, 35-36 Random effects analysis shows that the incidence of steroid-resistant rejection was significantly lower in the experimental group (RR 0.66; CI 0.48-0.91; P = 0.011; six trials; Fig. 3, Table 3). Meta-regression and subgroup analysis showed that the effect is evident only in randomized trials (RR 0.65; CI 0.47-0.91; P = 0.011; five trials) and at three to six months (RR 0.44; CI 0.21-0.94; P = 0.03; three trials), but not at 12 months or later. Furthermore, we did not observe other significant covariates and there was no significant heterogeneity in any of the analyses (Table 4).

Figure 3.

Forest plot of steroid-resistant acute rejection. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.].

Table 4. Additional results of meta-analysis
Subcategory & OutcomeNo of trialsNo of analysed patientsWeighted mean difference [95% CI]Test for overall effectTests for heterogeneity
z valueP valueP valueI2 (%) [95% CI]
  • Relative risk values <1 favour treatment with IL-2Ra. For estimated glomerular filtration rate (aGFR) weighted mean difference values >0 favour treatment with IL-2Ra whereas for serum creatinine values <0 favour IL-2Ra. Abbreviations: CI, confidence interval; HCC, hepatocellular carcinoma; NA, not available; PTDM, post-transplant diabetes mellitus; PTLD, post-transplant lymphoproliferative disorder.

  • *

    Data only available for studies in comparison 2.

Renal dysfunction
eGFR (mL/min), 12 months or later, only comparison 2*36416.29 [1.66 to 10.91]2.670.0080.750 [0 to 63]
S-creatinine (mg/dl), 3 to 12 months, all comparisons615800.01 [−0.07 to 0.08]0.140.890.00868 [15 to 21]
S-creatinine (mg/dl), 6 to 12 months, only comparison 23580−0.05 [−0.11 to −0.01]2.190.030.410 [0 to 98]
Relative risk [95% CI]
Renal dysfunction, all camparisons614760.55 [0.32 to 0.96]2.120.030.2031 [0 to 91]
Renal dysfunction, only comparison 257780.46 [0.27 to 0.78]2.870.0040.3510 [0 to 91]
Steroid resistant rejection
Three months27460.44 [0.21 to 0.94]2.110.030.64NA
One year or later38450.74 [0.48 to 1.15]1.340.180.64NA
Overall515910.66 [0.48 to 0.91]2.550.010.550 [0 to 86]
Infection
All cause1121401.00 [0.88 to 1.14]0.010.990.190 [0 to 95]
Cytomegalovirus infection817750.85 [0.55 to 1.32]0.720.470.1246 [0 to 90]
Malignancy
All (except recurrence of HCC)712210.85 [0.38 to 1.91]0.400.690.780 [0 to 67]
Lymphoma/PTLD33120.94 [0.16 to 5.33]0.070.940.990 [NA]
Metabolic and cardiovascular disorders
Hyperlipoproteinemia516490.92 [0.62 to 1.37]0.390.700.500 [0 to 94]
Hypertension, all comparisons717490.95 [0.80 to 1.13]0.540.590.4511 [0 to 78]
Hypertension, only comparison 3512200.87 [0.68 to 1.10]1.150.250.476 [0 to 84]
PTDM, all comparisons1020590.56 [0.39 to 0.82]3.030.0020.0158 [6 to 83]
PTDM, only comparison 3613860.41 [0.30 to 0.55]5.790.0010.880 [0 to 59]
All adverse events and serious adverse events
Adverse events314171.00 [0.99 to 1.01]0.150.880.890 [0 to 82]
Serious adverse events27191.05 [0.95 to 1.15]0.940.350.790 [0 to 99]

Graft Loss and Patient Death.

Fifteen studies23-24, 27-38, 40 reported data on graft loss and all but one trial39 reported data on patient death. Random effects analysis showed insignificant effects for both graft loss (Fig. 4) and patient death (Fig. 5). In the meta-regression (Table 3) we only found measurements of graft loss at 12 months or later to be significantly lower than measurements at 3-6 months (ratio of RR 0.60; CI 0.38-0.94; P = 0.03), but subgroup analysis did not show significant effects in either subgroup. There was only marginal heterogeneity in all analyses and funnel plot analysis suggested only missing studies at RR higher than one.

Figure 4.

Forest plot of graft loss stratified by randomization status of the studies. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.].

Figure 5.

Forest plot of patient death stratified by randomization status of the studies. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.].

Renal Function

In six trials,24, 26, 31, 34, 39-40 immunosuppression with IL-2Ra in combination with delayed or reduced CNI was compared to standard immunosuppression without IL-2Ra (comparison 2). The rationale of avoiding early and standard dose CNI is to reduce the adverse effects of CNI, especially nephrotoxicity.43 In the subgroup of comparison 2 we therefore planned to analyze mid- to long-term renal function by comparing the appropriate surrogate markers, i.e., serum creatinine and/or eGFR. Of the six trials two reported only the incidence of renal dysfunction but not eGFR or creatinine,26, 39 another two reported only either eGFR or creatinine,31, 40 and two trials reported both.18, 34 GFR was either estimated by Cockcroft-Gault44 or the MDRD formula.45 In studies included in comparison 2, both the analysis of eGFR and of serum creatinine favored the use of IL-2Ra (see Table 4). The analysis of serum creatinine in all comparisons revealed no difference in effect but significant heterogeneity (P = 0.008). Sources of heterogeneity were explored by meta-regression and we found that opposing effects in studies from comparison 2 and 3 caused considerable heterogeneity (P = 0.007).

Complications and Side Effects of Immunosuppression

A limited number of trials reported data on complications, side effects, and (serious) adverse events (AE/SAE). We found no differences in the incidence of infection, malignancy, and overall AE/SAE (Table 4). Posttransplant diabetes mellitus (PTDM) was less common in patients treated with IL2-Ra (RR 0.56; CI 0.39-0.82; P = 0.002; 10 studies). Heterogeneity observed in this analysis is mainly due to opposite effects in the different comparison groups (P < 0.001) with significant reduction of PTDM mainly in comparison 3 (RR 0.41; CI 0.30-0.55; P = 0.001; six studies/cohorts) compared to the other comparison groups. We found no other significant differences in metabolic and cardiovascular disorders.

The analysis of renal dysfunction showed a clear benefit for IL-2Ra when combined with low-dose and/or delayed CNI (RR 0.46; CI 0.27-0.78; P = 0.004; five trials) and we found this effect also in the joint analysis of all comparisons (RR 0.55; CI 0.32-0.96; six trials; P = 0.03).

Only three trials23, 29, 38 reported the occurrence of hepatitis or fibrosis after liver transplantation, but the data could not be pooled because definitions of how the diagnosis was obtained were lacking. Only one trial reported detailed data on recurrence of fibrosis in patients with HCV infection,29 but there were no significant difference between the two treatment arms.

Discussion

The use of IL-2Ra in addition to standard double-drug or triple-drug therapy significantly reduces the risk of acute rejection and steroid-resistant rejection after liver transplantation. Subgroup analysis and meta-regression for acute rejection showed that this effect is more prominent in studies that included MMF as concomitant immunosuppression (in both groups) but independent of other study-level covariates such as type of IL-2Ra and type of CNI. IL-2Ra also seem to be more effective in studies included in comparison 2, but all of these studies also included MMF and after adjusting for use of MMF this effect is no longer seen. Acute rejection was significantly reduced only at 12 months or later, whereas steroid-resistant acute rejection was significantly reduced only in the analysis of trials that assessed rejection at 3 months. Hence, it remains unclear whether the effect of IL2-Ra may attenuate over time.

Although the risk of acute rejection is reduced when IL2-Ra are applied, we did not observe a significant reduction in graft loss or patient death. Observed trends suggested that the patient collective may be too small to observe significant effects. On the other hand, the correlation between acute rejection and graft loss after liver transplantation may not be as strong as implicated.12

We also looked at the possibility of reducing concomitant immunosuppressive medication when using IL-2Ra because most published studies explored this effect. In patients receiving low-dose and/or delayed CNI in combination with IL-2Ra (comparison 2) we observed significantly better long-term renal function both with regard to serum creatinine and estimated GFR than in patients with standard immunosuppression. In this group we also observed a significant reduction in the incidence of renal dysfunction. Furthermore, patients with early discontinuation of steroids or steroid avoidance in combination with IL-2Ra (comparison 3) had a significantly lower incidence of PTDM.

The analysis of all included trials showed that therapy with IL-2Ra was not associated with an increased incidence of malignancies, bacterial or viral infection, or adverse events in general, indicating that IL2-Ra are safe and without significant side effects for at least 12 months after liver transplantation. Longer follow-up has been reported for registry data and corroborates this analysis.46

Strength and Limitations

The main limitation of this review is the low number of randomized controlled trials, even compared to kidney transplantation,7 which makes it difficult to acquire enough data for meaningful results. After a first unsystematic review we decided to include not only randomized trials but also nonrandomized controlled trials in this review. Very few studies only compared IL-2Ra to placebo or no treatment and many more studies explored the effects of reduced or delayed concomitant immunosuppression. Therefore, we decided to include those studies in the analysis by allocating them to predefined comparisons. Furthermore, we included and pooled studies that used a different type of IL2-Ra, had different concomitant medication (CNI and MMF), or had different follow-up times.

Because all these differences may be sources of heterogeneity, it was planned to perform joint analyses and also to explore differences of effect by performing subgroup analyses and meta-regression. Due to the paucity of data on secondary outcomes we were only able to extensively analyze the primary endpoints.

Another problem was the insufficient detailed reporting of outcomes; this was most evident regarding the side effects of immunosuppression. Not only did few studies give data on complications and side effects, but also these were reported in insufficient detail or were measured or grouped differently in the various trials. We endeavored to overcome this limitation by grouping data on side effects into broader categories, but this may further limit the interpretation of the results.

External and internal validity of the trials and the results of this meta-analysis are difficult to assess because important methodological details were omitted in the trial reports. Although we attempted to minimize publication bias by searching for and including data from different databases, conference abstracts, and non-English language sources, the inclusion of such data further hindered assessment of validity.

Nonetheless, this review and meta-analysis was conducted at an appropriate time because enough data has accumulated for a first inspection by meta-analytical methods. We do not expect more data to accumulate over the next years unless further trials are demanded of the proprietors of commercially available IL-2Ra preparations by the regulatory authorities.

Clinical Implications

Fifteen patients would need to be treated to prevent one acute rejection (NNT [number needed to treat] = 15) and 29 patients would need to be treated to prevent one steroid-resistant rejection (NNT = 29). The risk reduction for acute rejection is less than would be expected from experience with kidney transplantation7 but because the absolute risk of rejection is lower in liver transplantation this is a reasonable result. The numbers are comparable with regard to steroid-resistant rejection. Basiliximab and daclizumab seem to be equally effective in reducing the risk of rejection. In protocols with steroid avoidance (comparison 3), eight patients would need to be treated to prevent one event of PTDM.

In conclusion, the use of IL-2Ra reduces the risk of acute rejection and steroid-resistant acute rejection without an increase of harmful effects. This effect allows for reduction of coimmunosuppression to avoid the adverse side effects of CNI or steroids. Harnessing this immunological umbrella may enable patient-tailored immunosuppression such as low-dose, delayed CNI for the patient at risk for renal failure or steroid avoidance for patients at risk for PTDM and other metabolic side effects of steroids. The beneficial effects of IL-2Ra should be further evaluated in the context of comparative effectiveness research.

Acknowledgements

We thank Prof. Tim Friede (Department of Medical Statistics, University Medical Center Göttingen) for reviewing the article and for statistical advice. We would also like to thank the three reviewers of the article as their critical and helpful comments allowed us to substantially improve the publication.

Ancillary