Liver Transplant Program/Multi-Organ Transplant Program, University Health Network/Toronto General Hospital, University of Toronto, Toronto, Ontario, Canada
Liver Transplant Program/Multi-Organ Transplant Program, University Health Network/Toronto General Hospital, University of Toronto, 585 University Avenue, NCSB 11C-1238, Toronto, Ontario M5G 2N2, Canada
Eberhard L. Renner has served as a member of advisory boards for Astellas, Novartis, Roche, and Vertex; as a member of the scientific committee for the SUSTAIN multicenter clinical trial (Novartis); and as a speaker for Novartis. He has also received grant support from Novartis Canada and Roche Canada.
This work was supported by an unrestricted grant to Eberhard L. Renner from Novartis Canada.
Chronic hepatitis C virus (HCV) infection is the leading indication for liver transplantation (LT) worldwide.1, 2 Although LT offers the optimal therapy for HCV-related end-stage liver disease and early-stage hepatoma, universal HCV reinfection of the graft is a major concern. The course of recurrent HCV infections after transplantation is accelerated, and recurrence results in graft cirrhosis in up to 30% of recipients within 5 years. Long-term graft survival and patient survival are, therefore, impaired.3
The successful treatment of recurrent HCV, which is demonstrated by sustained HCV clearance or a sustained virological response (SVR), is associated with reduced liver-related mortality4 and improved overall survival4, 5; a combination of pegylated interferon-α (PEG-IFNα) and ribavirin (RBV) is the current standard of care. Treatment efficacy after transplantation remains, however, inferior to efficacy in the pretransplant setting, with SVR rates ranging from 15% to 40% for genotype 1 (G1)/genotype 4 (G4) patients and from 44% to 88% for genotype 2 (G2)/genotype 3 (G3) patients.4, 6, 7 Improving the efficacy of antiviral therapy (AVT) is, therefore, a pressing priority in transplant hepatology.
In vitro, cyclosporine A (CSA), but not tacrolimus (TAC), has been shown to dose-dependently inhibit HCV replication.8 In addition, CSA appears to augment the antiviral effect of interferon-α (IFNα) in vitro.9, 10 Although some recent clinical reports have suggested that AVT may be more effective in LT recipients treated for recurrent HCV who are on CSA-based immunosuppression versus TAC-based immunosuppression,4, 11, 12 the clinical evidence remains conflicting, and appropriately powered randomized controlled trials (RCTs) comparing the efficacy of AVT for recurrent HCV with CSA versus TAC are lacking.
We, therefore, conducted a systematic review and meta-analysis to test the hypothesis that the efficacy of AVT (ie, the SVR rate) is higher for HCV-positive LT recipients treated with CSA versus TAC.
AVT, antiviral therapy; CI, confidence interval; CSA, cyclosporine A; df, degrees of freedom; EOT, end of treatment; G1, genotype 1; G2, genotype 2; G3, genotype 3; G4, genotype 4; HCV, hepatitis C virus; IFN, interferon; IL, interleukin; LT, liver transplantation; M-H, Mantel-Haenszel; MMF, mycophenolate mofetil; NR, not reported; PC, prospective cohort; PEG-IFN, pegylated interferon; RBV, ribavirin; RC, retrospective cohort; RCT, randomized controlled trial; RR, relative risk; SE, standard error; SVR, sustained virological response; TAC, tacrolimus.
MATERIALS AND METHODS
A detailed set of eligibility criteria were established a priori to identify studies relevant for this systematic review (see the Study Selection section). A comprehensive search of published and unpublished RCTs and observational studies involving adult subjects (18 years old or older) who had received IFN-based combination AVT for recurrent HCV after LT was undertaken. An a priori decision was made to include only studies using combination therapy with IFN (PEG-IFN or standard IFN) and RBV for the treatment of established HCV recurrence because this is widely regarded as the current standard of care. Thus, reports of therapy based on only IFN (PEG-IFN or standard IFN) or RBV and studies of preemptive therapy were excluded. There were no language restrictions during our search process.
An electronic search of the following databases was performed: Ovid MEDLINE (January 1995 to May 2012), Embase (January 1995 to May 2012), Scopus (January 1995 to May 2012), Cochrane Database of Systematic Reviews (up to May 2012), Cochrane Central Register of Controlled Trials (up to May 2012), and Database of Abstracts of Reviews of Effects (January 1995 to May 2012). The search strategy was limited to humans and combined the following keywords: liver or hepatic; transplantation, transplant, or graft; hepatitis C, hep C, or HCV; and interferon, IFN, pegylated-interferon, or PEG-IFN. Micromedex13 was used to determine alternate drug names. The electronic search was supplemented with a manual search of original studies and conference abstracts (American Association for the Study of Liver Diseases, European Association for the Study of the Liver, International Liver Transplantation Society, and Digestive Disease Week) published in the journals Hepatology, Journal of Hepatology, Liver Transplantation, and Gastroenterology between January 2005 and May 2012. Reference lists of retrieved articles and other reviews were screened to identify other potential studies. When it was necessary, we attempted to contact researchers to identify missing data not included in the original publication.
Titles and abstracts identified during the search selection process were independently screened by 2 authors (R.R. and E.L.R.). Pilot testing was employed for the first 50 citations to ensure explicit and valid screening criteria and good (>80%) interobserver agreement. All full-text articles from potentially eligible studies were then retrieved and independently reviewed by 2 authors (R.R. and E.L.R.) with strict inclusion and exclusion criteria. We included controlled studies and observational studies published as full texts or abstracts in which standard IFN and/or PEG-IFN plus RBV were used for the treatment of established recurrent HCV (which was diagnosed by histology, biochemistry, and the detection of HCV RNA in serum). Combination therapy had to be planned for at least 24 weeks, the studies had to report on the primary outcome (SVR), and data on the immunosuppressive regimen at the time of AVT initiation had to be clearly stated. We excluded the following: (1) companion reports (studies with fewer patients and/or shorter follow-up were excluded); (2) electronic search–identified abstracts published before 2005; (3) review articles, editorials, and case reports; (4) small studies or case series with less than 10 patients on CSA or TAC; (5) studies of non-LT or multiorgan transplant patient populations; (6) studies including only patients coinfected with human immunodeficiency virus or hepatitis B; (7) studies of patients receiving AVT for HCV after retransplantation; (8) studies in which not all treated patients received combination therapy or therapy was used in conjunction with other/experimental interventions or treatments; (9) studies in which calcineurin inhibitor–specific SVR rates could not be determined; and (10) studies strictly reporting highly selected patient populations only (eg, patients undergoing retreatment with IFN-based combination therapy after a previous post-LT course of AVT had failed). When more than 1 exclusion criterion existed, the primary reason (ie, the exclusion criterion first encountered on the aforementioned list) was selected. Any disagreements were resolved by discussion and in consensus with the third author (K.M.).
Data Collection Process
A data collection form was designed and piloted by 2 authors independently (R.R. and K.M.) and was then shared to ensure that any questions about data items were answered before full data extraction. Both reviewers then entered data independently for the primary and secondary outcomes, AVT treatment characteristics, and patient demographics. Discrepancies were resolved by a re-review of the original publication, discussion, and consensus.
All studies were also evaluated independently by 2 reviewers (R.R. and K.M.) for methodological quality with the Newcastle-Ottawa Scale for cohort studies.14 This scale awards a maximum of 9 points (range = 0-9) for good-quality studies with a low risk of bias. Again, discrepancies were resolved by a re-review of the original publication, discussion, and consensus. To assess publication bias, a funnel plot of the included trials was constructed.
Data Analysis and Statistics
Primary and Secondary Outcomes
The primary outcomes were the pooled SVR rates (for all genotypes) with IFN-based combination therapy for recurrent HCV with CSA and TAC; pooled genotype-specific SVR rates (G1 or G1/G4) according to the calcineurin inhibitor were also analyzed. The secondary outcomes were the pooled end-of-treatment (EOT) response rates and the pooled relapse rates (all genotypes).
We used 3 prespecified stratifying variables to conduct subgroup analyses: (1) full-text publications, (2) studies using a combination of PEG-IFN and RBV in all patients (ie, studies in which some patients were treated with a standard IFN/RBV combination were excluded), and (3) larger studies with 40 or more patients treated with each drug (CSA and TAC).
For dichotomous outcomes, relative risks (RRs) and 95% confidence intervals (CIs) were calculated for individual studies, and the summary statistics were calculated with a random effects model (Review Manager version 5.0, Nordic Cochrane Centre, Copenhagen, Denmark). For all statistics, a P value ≤ 0.05 was regarded as statistically significant. Heterogeneity was analyzed with Cochran's Q statistic (α = 0.05 for statistical significance) and with the I2 statistic, which is derived from Q and describes the proportion of total variation that is due to heterogeneity beyond chance.15 We planned to perform a multivariate meta-regression analysis, but we were not able to do so because of inconsistencies in the data reporting of the included studies.
Our initial electronic database search yielded 5058 citations; 2614 were independent (2444 citations were duplicates of identical citations resulting from searches of different databases; see Fig. 1). We excluded 2389 citations after title and abstract screening because they clearly did not fulfill our eligibility criteria. The κ agreement between the 2 reviewers was 0.843 ± 0.020 at this stage, which indicated very good interobserver agreement. After a detailed full-text review of 225 references, 18 studies, including 14 full-text articles and 4 abstracts, were included in the systematic review.4, 5, 7, 11, 12, 16-28 The interobserver agreement at this stage was 1.00. Because of inherent differences in the study design, the single RCT19 was excluded from the meta-analysis portion of our study, which comprised the observational studies only. A funnel plot had the appearance of a relatively symmetric, inverted funnel (Fig. 2), which suggested little or no risk of publication bias.
A description of the baseline features of the included studies is provided in Table 1. The studies were conducted primarily in Europe (n = 13),5, 7, 11, 12, 16, 17, 22-28 with the remainder originating from North America (n = 4)4, 18-20 and Japan (n = 1)21; all were published between 2006 and 2012. All but 1 study19 were uncontrolled with prospective (n = 6)5, 7, 16, 20, 23, 24 or retrospective designs (n = 11).4, 11, 12, 17, 18, 21, 22, 25-28 The study by Carrión et al.17 was a randomized, open-label, controlled study (AVT versus no treatment); however, we used data only from the treated study arm and, therefore, considered this an uncontrolled, prospective design for our analysis. All full-text publications defined eligibility criteria and post-LT AVT in detail, but they were less well described in the 4 abstracts.16, 20, 26, 28 In general, however, the diagnosis of recurrent HCV infections was uniform across all studies and consisted of a combination of detectable HCV RNA in serum (polymerase chain reaction), elevated serum transaminases, and evidence of histological changes in liver biopsy samples compatible with recurrent HCV infections. Nine studies4, 7, 11, 18, 19, 23-25, 27 provided the pre-LT AVT history, and they demonstrated that the majority of the patients were IFN-naive (range = 62%-100%). The majority of the studies did not explicitly state that the LT procedures were cadaveric in nature, but according to those studies providing this information (n = 5), deceased donor LT was performed for 85% to 100% of the participants.4, 11, 12, 19, 20 The exception was a single trial reporting on living donor LT recipients only.21 In all the studies, the majority of the participants were male (range = 60%-85%) with G1 disease (range = 66%-100%), whereas the interval between LT and the initiation of AVT was quite variable (range = 1-231 months).
Table 1. Characteristics of the Studies Included in the Systematic Review and Meta-Analysis
Design/Country/ Center Type
All Evaluable Cases (n)
Male Sex (%)
Interval From LT to Treatment (Months)
Target Combination Treatment Duration (Weeks)
SVR [% (n/N)]
NOTE: The meta-analysis was performed with observational studies; a single RCT19 was not included in the meta-analysis.
Pilot study comparing SVR with CSA versus SVR with TAC
TAC: 35 (7/20)
No significant difference in SVR with CSA or TAC
Eleven studies, including the single RCT, reported on IFN-based antiviral combination therapy using PEG-IFN (α2a and/or α2b) and RBV exclusively,5, 7, 11, 16, 17, 19, 20, 22-24, 27 whereas in 7 studies, standard IFN and PEG-IFN were each used in a proportion of patients.4, 12, 18, 21, 25, 26, 28 All full-text publications detailed a treatment schedule including changes in the doses of IFN and RBV (as tolerated by the participants). Thirteen studies mentioned the use of erythropoietin and granulocyte colony-stimulating factor in patients with anemia and leukopenia, respectively.4, 7, 11, 12, 16-19, 22-25, 27 The intended duration of therapy was 48 weeks regardless of the genotype (n = 8),4, 7, 17, 20, 25-28 was 52 weeks (n = 1),16 or was dependent on virological responses (n = 2).21, 23 Six studies planned 24 weeks of treatment for G2/G3 patients and 48 weeks for G1 patients.5, 12, 18, 19, 22, 24 The rate of adherence to the full dose and duration of therapy for both IFN/PEG-IFN and RBV was 30% to 50% according to full-text articles reporting this information (n = 10).4, 5, 7, 17-20, 22, 23, 25 Tolerability as assessed by the 80/80/80 rule (the ability to tolerate 80% of the recommended doses of RBV and IFN/PEG-IFN at least 80% of the time) was reported by 7 studies and ranged from 31% to 62%.4, 7, 11, 18, 20, 25, 27
All publications included patients treated with either CSA or TAC as the primary immunosuppressive agent at the initiation of AVT; the criteria for using CSA or TAC in an individual patient were not clearly stated in the observational reports. Concomitant immunosuppression was highly variable across the studies, with all full-text articles providing some description of the strategy used. In general, efforts were made to taper off steroids within 3 to 6 months of LT, and in 8 studies, mycophenolate mofetil (MMF) was used in conjunction with the calcineurin inhibitor.4, 5, 11, 12, 20, 21, 24, 27
Quality of the Included Trials
The study quality was rated independently by 2 reviewers (R.R. and K.M.) who used the Newcastle-Ottawa Scale for cohort studies.14 All observational studies (full texts and abstracts) adhered to the majority of the guidelines outlined in the scale, and the mean score was 7.5 (range = 6-9), with 9 studies receiving ratings of 8 or 9 out of a maximum of 9.4, 5, 11, 12, 18, 25-28 Eleven cohort studies4, 5, 11, 12, 18, 20, 21, 23, 25-27 indicated that patients were consecutively sampled for study enrollment, with the remaining studies using a selective sampling scheme (n = 4)7, 17, 22, 24 or not explicitly describing the methods of patient selection (n = 2)16, 28; those without consecutive sampling were deemed to have at least a moderate risk of bias.
Primary Outcome: SVR With CSA and TAC
By virtue of our selection criteria, the SVR rates were reported in all 18 studies. The sole RCT19 found a numerically higher SVR rate with CSA (39%) versus TAC (35%), but the difference was too small to reach statistical significance because of the limited power of this small pilot-type trial. The pooled SVR rates (for all genotypes) from all 17 observational studies were 42% (395/945 patients) with CSA and 35% (471/1364 patients) with TAC. This compares favorably with the SVR rates reported in the aforementioned RCT. A meta-analysis using a random effects model confirmed higher efficacy for AVT with CSA versus TAC (RR = 1.18, 95% CI = 1.00-1.39, P = 0.05), but there was significant heterogeneity across the studies (I2 = 45%, P = 0.02; Fig. 3A).
Genotype-specific SVR rates according to the calcineurin inhibitor were available from only 4 studies4, 18-20; all of these studies reported higher SVR rates for G1 patients on CSA versus G1 patients on TAC. The RCT by Firpi et al.19 reported numerically (albeit not significantly) higher SVR rates for G1 patients (n = 17 for each arm): 35% with CSA and 31% with TAC. Selzner et al.4 identified SVR rates for HCV G1/G4–infected patients, which were higher with CSA versus TAC (49% versus 34%, P = 0.01); the small number of G2/G3 patients precluded a similar analysis for that subgroup. Firpi et al.18 discovered higher SVR rates for HCV G1–infected patients on CSA versus TAC [44% versus 17% (not statistically analyzed)] but numerically (albeit not significantly) lower SVR rates for the small number of HCV G2–infected patients (67% versus 80%) and HCV G3–infected patients (40% versus 80%). Similarly, Gordon et al.20 reported higher SVR rates for HCV G1–infected patients [33% versus 24% (not statistically analyzed)]; only 2 G2/G3 patients were on CSA, so comparisons for non-G1 infections were limited. A pooled analysis of the 3 observational studies showed a significantly higher SVR rate for G1 patients18, 20 and G1/G4 patients4 on CSA versus TAC (RR = 1.64, 95% CI = 1.15-2.34, P = 0.007; Fig. 3B). The study heterogeneity was low for this G1- and G1/G4-specific analysis (I2 = 6%, P = 0.34).
Only 1 study aimed to examine the interaction between donor and recipient interleukin-28B (IL-28B) genotypes and calcineurin inhibitor–specific SVR rates.28 Bitetto et al.28 identified a significant association between recipient IL-28B genetic polymorphisms and SVR in CSA-treated patients (but not TAC-treated patients).
Secondary Outcomes: EOT Response and Relapse Rates With CSA and TAC
Calcineurin inhibitor–specific EOT response and relapse rates were reported in 6 studies (Table 2).4, 11, 12, 18, 19, 26 The pooled EOT response rates for the observational studies were 63% (304/479) with CSA and 60% (456/762) with TAC, and they were not significantly different (RR = 1.13, 95% CI = 0.94-1.36, P = 0.19); however, the relapse rates were lower with CSA (19%) versus TAC (26%; RR = 0.77, 95% CI = 0.61-0.96, P = 0.02). The study heterogeneity for the pooled analyses was moderate to high for the EOT response rates (I2 = 69%, P = 0.01) and the relapse rates (I2 = 73%, P = 0.005).
Table 2. EOT Response Rates and Relapse Rates by the Calcineurin Inhibitor in Studies With Evaluable Data
Analyzing only the 13 full-text observational publications (with 1445 patients in all) yielded a numerically (albeit not significantly) higher SVR rate with CSA versus TAC (RR = 1.14, 95% CI = 0.91-1.44, P = 0.26)4, 5, 7, 11, 12, 17, 18, 21-25, 27; the heterogeneity was moderate to high (I2 = 57%, P = 0.006; Fig. 4A).
Limiting the analysis to the 10 observational studies using PEG-IFN with RBV only (with 1095 patients in all) resulted in an SVR rate with CSA that was numerically (but not significantly) higher than the rate with TAC (RR = 1.10, 95% CI = 0.87-1.40, P = 0.43).5, 7, 11, 16, 17, 20, 22-24, 27 The study heterogeneity remained moderate, although it was no longer statistically significant (I2 = 38%, P = 0.11; Fig. 4B). When the analysis was confined to full-text studies using PEG-IFN and RBV exclusively,5, 7, 11, 17, 22-24, 27 the results (RR = 1.05, 95% CI = 0.78-1.42, P = 0.75) and the study heterogeneity (I2 = 47%, P = 0.07) were not significantly altered.
Restricting the analysis to the 7 observational studies with 40 or more patients in each group (with 1634 patients in all) yielded a higher SVR rate with CSA versus TAC (RR = 1.23, 95% CI = 1.09-1.38, P < 0.001).4, 11, 18, 25-28 In this analysis of large studies, heterogeneity was essentially eliminated (I2 = 0%, P = 0.62; Fig. 4C). The exclusion of studies published only in abstract form26, 28 had no relevant impact on the results (RR = 1.23, 95% CI = 1.06-1.43, P = 0.006) or the study heterogeneity (I2 = 0%, P = 0.42).
Other Outcomes: CSA Versus TAC
The development of acute or chronic cellular rejection during AVT was a low-occurrence event (0%-18%) in all the full-text studies except for the 2 studies published by the same group,18, 19 which reported a higher number of rejection episodes (Table 3; the data were not provided in the abstracts). Calcineurin inhibitor–specific rejection rates were evaluable from 4 cohort studies and the RCT and are presented in Table 3.7, 11, 12, 18, 19 The rejection rates were numerically higher for the CSA-treated patients in all the studies except for the largest full-text study, which identified identical rates of 5% in both groups.11 Because of the variability in (1) the criteria used for diagnosing/defining rejection episodes, (2) the target CSA/TAC levels, and (3) therapeutic drug monitoring (CSA trough levels versus levels 2 hours after ingestion), the results allowed several interpretations and were, therefore, not pooled.
Table 3. Rejection Rates and Episodes With CSA and TAC During AVT
Rejection Rate [% (n/N)]
Rejection episodes were reported; the number of events per patient was unknown.
Three studies reported on the calcineurin inhibitor–specific tolerability of AVT and the need for the discontinuation of therapy.12, 18, 19 The discontinuation rates before the completion of the intended duration of AVT ranged from 11% to 17% for CSA and from 8% to 24% for TAC. The RCT described greater PEG-IFN dose reductions in the CSA group versus the TAC group (94% versus 60%).19 Additionally, 1 cohort study identified an independent association between CSA-based immunosuppression and the development of anemia; anemia itself was, however, not associated with a poor AVT response.27
Patient and Graft Survival
Patient survival was assessed according to the calcineurin inhibitor in 3 studies.12, 18, 26 Firpi et al.18 reported overall death rates to be higher for the CSA-treated patients versus the TAC-treated patients (23% versus 10%), but these rates were unadjusted for the follow-up duration, which was significantly longer for the CSA group. No difference in patient survival was identified by Vero et al.26 at a median follow-up of 54.2 months (88% survival with CSA and 87% survival with TAC). Cescon et al.12 provided 5-year patient and graft survival rates after LT and reported statistically insignificant differences between CSA- and TAC-treated patients (5-year patient survival, 79% with CSA versus 68% with TAC; 5-year graft survival, 79% with CSA versus 66% with TAC).
Our study is the first to systematically review the literature on the efficacy of AVT for recurrent HCV after LT in patients treated with CSA- or TAC-based immunosuppressive regimens. Our meta-analysis of 17 studies with a total of 2309 patients identified a borderline statistically significantly higher SVR rate with AVT and CSA (pooled SVR rate = 42%) versus AVT and TAC (pooled SVR rate = 35%; RR = 1.18, 95% CI = 1.00-1.39, P = 0.05), but moderate heterogeneity was present (I2 = 45%, P = 0.02). An analysis of the studies reporting calcineurin inhibitor–specific SVR rates for G1 patients and G1/G4 patients demonstrated a significant benefit from AVT with CSA (RR = 1.64, 95% CI = 1.15-2.34, P = 0.007) with low heterogeneity across the studies (I2 = 6%, P = 0.34). In a planned subgroup analysis of the larger studies only (≥40 patients in each group), SVR rates remained statistically higher with CSA versus TAC (RR = 1.23, 95% CI = 1.09-1.38, P < 0.001), and the study heterogeneity was eliminated (I2 = 0%, P = 0.62).
Two published studies have suggested a reduction in the relapse rate as the primary explanation for the improved SVR rates with CSA.4, 11 Our meta-analysis confirms a significantly lower relapse rate for CSA-treated patients versus TAC-treated patients (RR = 0.77, 95% CI = 0.61-0.96, P = 0.02).
The beneficial effects of CSA may be limited to patients with HCV G1. However, genotype-specific data were available from only a minority of the studies, and the number of reported non-G1 patients was too small to allow a meaningful analysis. Thus, firm conclusions cannot be drawn. Similarly, the reporting of rejection rates and longer term patient and graft survival was subject to variability, and these data were inconsistently reported across the studies; this precluded meaningful comparisons.
The mechanism or mechanisms by which CSA improves the efficacy of IFN-based AVT in LT recipients with recurrent HCV are not fully understood. They may, however, include direct antiviral properties, with CSA (but not TAC) having been shown to inhibit HCV replication8, 18, 29 and to add to the antiviral efficacy of IFN in vitro.9, 30 It has furthermore been shown that HCV RNA polymerase (nonstructural protein 5B) requires binding to cyclophilin B to reach full activity. CSA, but not TAC, binds to cyclophilins, including cyclophilin B, and thereby inhibits the latter's binding to HCV RNA polymerase, polymerase activity, and HCV RNA synthesis.31, 32 Insulin resistance may also play a role. Thus, diabetes has been shown to decrease SVR rates for patients on AVT after LT,32 and interestingly, de novo diabetes after LT has been reported to be more frequent for patients on TAC versus patients on CSA.33, 34
Recent observations by Bitetto et al.28 are intriguing and suggest a unique interplay between CSA and recipient IL-28B genotypes in the achievement of SVR after LT. Their discovery that recipient IL-28B genetic polymorphisms are significantly associated with SVR in CSA-treated patients (but not TAC-treated patients) warrants confirmation and further study. However, the impact of donor IL-28B genotypes, which was not reported in the study by Bitetto et al., will also need to be considered because of recent data suggesting a predictive relationship between both donor and recipient IL-28B polymorphisms and SVR after LT.35
In contrast to the pretransplant setting, large RCTs of AVT for HCV recurrence after LT are lacking. We, therefore, had to rely on retrospective cohort (RC) and prospective cohort (PC) studies for our meta-analysis. We are aware of the inherent biases that exist in observational studies, and we attempted to partially overcome this by developing stringent eligibility criteria and by restricting our analysis to studies reporting an a priori defined minimum number of patients. Our inability to adjust for important differences in baseline characteristics that may have existed in CSA-treated patients and TAC-treated patients, however, remains a limitation of our analysis. Although our search criteria included studies published from 1995 onward, our requirement that studies had to report on certain a priori defined data elements resulted in the inclusion of only studies published from 2006 onward. The study heterogeneity was moderate in the pooled analysis of all included studies and likely reflected the changing practices in AVT over the time period of the study. However, the study heterogeneity was eliminated when the analysis was restricted to larger reports, which were also more recent, with all but 1 of these larger studies having been published between 2009 and 2012.4, 11, 25-28 We made an a priori decision to include unpublished data (abstracts) that met the eligibility criteria, although we recognized the limited methodological information provided. The literature on AVT after LT for recurrent HCV is, however, rapidly evolving, and we felt that it was important to capture the most up-to-date data. We did explore the possibility of a publication bias with a funnel plot; the clustering of the larger and more precise studies around the pooled treatment effect, with the less precise studies being symmetrically situated around it, suggests that a publication bias is unlikely to have relevantly affected our results.
Our study was not equipped to assess important predictive factors for SVR such as IL-28B polymorphisms36 because only 1 study had examined IL-28B genotypes.28 There is, however, no reason to believe that IL-28B distributions would have systematically differed between CSA-treated patients and TAC-treated patients in the various reports included in our analysis. Whether the marginally favorable effect of CSA will persist when regimens containing direct acting antivirals are employed remains to be seen. It is conceivable that in the post-LT setting, the decision to use a particular calcineurin inhibitor with these new regimens will be primarily dictated by the potential for drug interactions. It is interesting to note in this context that both telaprevir and boceprevir dramatically increased the exposure to TAC in recently published studies of healthy volunteers, whereas CSA exposure was affected to a lesser extent.37, 38
Although our results lend further support to a beneficial effect of CSA in patients with recurrent HCV who are undergoing AVT, they are not conclusive with respect to an overall benefit of using one or another calcineurin inhibitor as the backbone of immunosuppression in HCV-positive LT recipients. Thus, a recent United Network for Organ Sharing/Organ Procurement and Transplantation Network database study by Irish et al.39 identified marginally but significantly higher graft failure and patient death rates in CSA-treated patients. However, AVT was not assessed in that study, and its limitations included a small proportion of patients on CSA (8.1%) and potential cohort and center effects, which cast doubt on the comparability of CSA-treated patients and TAC-treated patients.
In conclusion, our systematic review identified a marginally higher RR for achieving sustained virological clearance with AVT and CSA versus AVT and TAC. Confirmation of this marginal benefit in an RCT is needed, although our results suggest that a very large study would be required to achieve the statistical power needed to definitively address this issue.
The authors thank the following for their invaluable assistance: Ani Orchanian-Cheff (information specialist), Kevin Thorpe (statistician), and Andrea Tricco (epidemiologist).