This study was supported by the National Science and Technology Pillar Program during the 11th 5-year planning period (Key Projects 2008BAI60B02 and 2008BAI60B06) and by the Key Clinical Project from the Ministry of Health (2010159) and Science and Technology Planning Project of Guangdong Province (Key Clinical Project 2008B030301308).
Hepatocellular carcinoma (HCC) is the most common primary malignancy of the liver. As the third leading cause of cancer-related deaths worldwide,1 HCC indicates an ominous prognosis for patients suffering from recurrence. For patients with unresectable HCC and particularly those with concomitant uncompensated cirrhosis, liver transplantation may be the only curative option.2 With the implementation of the Milan criteria (1 nodule ≤5 cm or 2 to 3 nodules with each ≤3 cm and no macroscopic vascular invasion or extrahepatic spread), the posttransplant survival of HCC patients has been greatly improved.3 Nevertheless, the risk of recurrence remains a serious problem.4
The use of immunosuppressants is essential for the prevention of graft rejection. However, immunosuppressants simultaneously contribute in part to tumor cell proliferation and metastasis. Calcineurin inhibitors (CNIs), which are the most commonly used immunosuppressants and include cyclosporine and tacrolimus, have been suggested to promote tumor growth in preclinical studies.5, 6 Both cyclosporine and tacrolimus have been reported to increase the posttransplant risk of HCC recurrence in a dose-dependent manner.7 Sirolimus (SRL), an inhibitor of the mammalian target of rapamycin (mTOR) pathway, is a novel immunosuppressant with antitumor effects.8 SRL can directly inhibit tumor cell proliferation by inhibiting mTOR or restraining vascular endothelial growth factor–mediated angiogenesis in established tumors.7-9 More importantly, both immunosuppressive and antitumor activity can be achieved with SRL at the target trough level in posttransplant patients.9, 10
In liver transplant recipients with a pretransplant diagnosis of HCC, retrospective studies have shown trends of lower recurrence and metastasis rates with SRL.11-14 However, most clinical trials in the literature have been conducted at single centers with small sample sizes and have not been assessed for possible confounding variables; the results of a multicenter, prospective, randomized controlled clinical trial are still pending.15 Thus, the efficacy of SRL is still a matter of debate. The aim of this study was to systematically summarize the currently available evidence for the efficacy and safety of SRL in liver transplant recipients with pretransplant HCC.
MATERIALS AND METHODS
The search was performed with MEDLINE, Embase, and the Cochrane Database of Systematic Reviews by 2 authors independently. Additionally, Google Scholar was used as a supplement. The Medical Subject Headings terms sirolimus and hepatocellularcarcinoma were used. To limit publication bias, no restriction by language or year was set in the search.
We included only studies evaluating the efficacy of SRL in patients undergoing liver transplantation for HCC. We required studies to compare SRL-based immunosuppression regimens to mTOR inhibitor–free immunosuppression regimens. Case reports were excluded.
We extracted data on overall survival, disease-free survival (DFS) or recurrence-free survival, recurrence rates, time to recurrence, acute rejection (AR) episodes, and adverse reactions to study drugs. The data were extracted by 2 investigators independently. The data collection and the assessment of the methodological quality were conducted as previously reported.16 The data collection and reporting were in accordance with the Quality of Reporting of Meta-Analyses statement.
The meta-analysis was performed according to the Cochrane Handbook for Systematic Reviews of Interventions,17 which is recommended by the Cochrane Collaboration. Pooled odds ratios (ORs) were calculated with fixed-effects models. The results were reported as pooled ORs and 95% confidence intervals (CIs). All calculations for the current meta-analysis were performed with Review Manager (version 5.0 for Windows, Cochrane Collaboration, Oxford, United Kingdom). This article follows the Quality of Reporting of Meta-Analyses and Cochrane Collaboration guidelines for reporting meta-analyses.
Assessment of Publication Bias and Heterogeneity
Graphical funnel plots were generated so that we could make visual inspections for publication bias.18 The statistical methods used for detecting funnel plot asymmetry were the Begg-Mazumdar rank correlation test19 and Egger et al.'s regression asymmetry test.18 Statistical heterogeneity was assessed with a forest plot and the inconsistency statistic (I2). Statistical significance was considered at P < 0.05.
AR, acute rejection; CI, confidence interval; CNI, calcineurin inhibitor; DFS, disease-free survival; HAT, hepatic artery thrombosis; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; I2, inconsistency statistic; MMF, mycophenolate mofetil; mTOR, mammalian target of rapamycin; N/A, not available; OR, odds ratio; SRL, sirolimus; M-H, Mantel-Haenszel.
As shown in the flow diagram (shown in Fig. 1), the use of the search terms sirolimus and hepatocellularcarcinoma revealed 92 articles. After the titles and abstracts were retrieved, 25 irrelevant or duplicate articles were excluded. Sixty-seven articles were reviewed in detail. Five controlled clinical trials with 2950 participants were included in the current meta-analysis.2, 20-23
Three retrospective cohort studies, a matched-cohort study, and a case-control study were included. The 5 studies involved 2 arms: an SRL-based protocol and an SRL-free protocol. Although not all patient enrollment was randomized in the included studies, no significant differences were observed with respect to the baseline variables. Table 1 summarizes the main characteristics of the included studies. Supporting Table 1 summarizes the characteristics of the studies that were excluded after their details were reviewed.
Table 1. Characteristics of the Randomized Trials Included in the Meta-Analysis
Three studies assessed the outcome measurements of 1-year patient survival (397 participants). The meta-analysis showed that in comparison with an SRL-free regimen, an SRL-based regimen improved 1-year patient survival (OR = 4.53, 95% CI = 2.31-8.89, P < 0.001). There was no significant heterogeneity between the studies (χ2 = 0.05, P = 0.97). Two studies reported 3-year patient survival (2718 participants). Both demonstrated survival benefits with an SRL-based regimen versus an SRL-free regimen. The meta-analysis indicated a significant improvement in 3-year patient survival with an SRL-based regimen versus an SRL-free regimen (OR = 1.97, 95% CI = 1.29-3.00, P = 0.002). There was no significant heterogeneity between the studies (χ2 = 2.37, P = 0.12). An analysis of 5-year survival (2815 participants) also showed significant benefits with an SRL-based regimen (OR = 2.47, 95% CI = 1.72-3.55, P < 0.001). Similarly, there was no significant heterogeneity between the studies (χ2 = 0.20, P = 0.90). Figure 2 summarizes the findings for the 1-, 3-, and 5-year survival of patients receiving SRL-based or SRL-free immunosuppression regimens.
Two studies (170 participants) reported 1-year DFS, and they showed the same trend of improved patient outcomes with SRL. The pooled results suggested benefits of SRL-based regimens versus SRL-free regimens (OR = 2.41, 95% CI = 1.10-5.30, P = 0.03). No significant heterogeneity was found (χ2 = 1.73, P = 0.19). Figure 3 shows the pooled results for the DFS of patients on SRL-based or SRL-free immunosuppression regimens.
Tumor Recurrence Rate
The tumor recurrence rates were reported for 3 trials. In all the trials, the tumor recurrence rates were lower for patients with an SRL-based regimen versus patients with an SRL-free regimen. The pooled analysis demonstrated a significant difference in the tumor recurrence rates that favored SRL-based regimens (OR = 0.42, 95% CI = 0.21-0.83, P = 0.01). There was no significant heterogeneity between studies (χ2 = 1.90, P = 0.39). Figure 4 shows the pooled results for the tumor recurrence rates of patients on SRL-based or SRL-free immunosuppression regimens.
Four studies reported the incidence of hepatic artery thrombosis (HAT), but studies by Zhou et al.2 and Vivarelli et al.21 reported no events in the follow-up period for the 2 groups. The results of these 2 studies were not estimable for the pooled outcomes, so they were discarded from the meta-analysis. No significant differences in the incidence of major SRL-related posttransplant complications such as acute cellular rejection (OR = 1.32, 95% CI = 0.86-2.03, P = 0.20) or HAT (OR = 2.36, 95% CI = 0.34-16.65, P = 0.39) were observed between SRL-based and SRL-free groups in this meta-analysis. Figures 5 and 6 show the pooled results for the AR episodes and HAT rates of patients on SRL-based or SRL-free immunosuppression regimens.
Heterogeneity and Publication Bias Assessment
No evidence of intertrial heterogeneity was found for the analyzed outcomes. As shown in Figs. 2 to 6, the P value of the heterogeneity assessment for each comparison was greater than 0.05 with I2 less than 25% [except for 3-year overall survival (P = 0.12, I2 = 58%) and DFS (P = 0.19, I2 = 42%)]. There was no evidence of publication bias, and the appearance of the funnel plot analysis was symmetrical (see Supporting Fig. 1). The Begg-Mazumdar test gave a P value of 0.73 for acute cellular rejection and a P value of 1.00 for other comparisons.
With respect to underlying causes of liver disease or cirrhosis, hepatitis B virus (HBV) was reported as a major cause only in the study by Zhou et al.,2 whereas hepatitis C virus (HCV) was the major cause in the included studies.20, 21, 23 The conclusions did not change when we excluded the study by Zhou et al.2 from comparisons of 1-year overall survival, tumor recurrence rates, and acute cellular rejection (Supporting Figs. 2-4). Similarly, when we excluded the results of Toso et al.20 from the 5-year overall survival analysis because of its large sample size, no different conclusion was generated (Supporting Fig. 5).
In this meta-analysis, we included data from 5 studies investigating the efficacy and safety of SRL in liver transplant patients for HCC. All pooled analyses of 1-, 3-, and 5-year patient survival showed a clear benefit from SRL-based regimens versus SRL-free regimens, and they indicated that SRL could provide better short- and long-term outcomes than CNIs. We found that all overall survival rates (eg, 2- and 4-year rates) in the studies (even those not included in this meta-analysis) favored the use of SRL. In addition, Zhou et al.2 reported a longer average survival time for patients with an SRL-based immunosuppression regimen versus an SRL-free immunosuppression regimen. Therefore, almost all the published studies supported the finding that SRL could extend the lifespan of liver transplant recipients with pretransplant HCC.
Because mTOR inhibition has been shown to be able to extend the lifespan of mice,24 it would be interesting to understand the mechanisms underlying the survival benefit of SRL in liver transplant recipients. To determine whether SRL could expand the lifespan of HCC patients by postponing death from cancer recurrence, we analyzed all data on tumor recurrence. The pooled results for 1-year DFS suggested benefits of SRL-based regimens versus SRL-free regimens. In addition, significant differences in the 3- and 5-year DFS rates between the groups were reported by Vivarelli et al.21 and Zimmerman et al.,22 respectively. Thus, we speculate that SRL improves both short- and long-term DFS. As for the tumor recurrence rate, only an overall tumor recurrence rate favoring the use of SRL could be found because of different follow-up times and incomplete records in the included studies. Moreover, we analyzed 2 studies that analyzed the time to recurrence.2, 21 However, no reliable conclusion could be drawn because the results of the studies were contradictory, and the level of heterogeneity was beyond the standard. Therefore, we could not conclude that the improved survival of the patients on SRL-based regimens was due to the antitumor properties of SRL.
More importantly, because none of the included studies performed a statistical analysis of the causes of death, we could not determine whether the survival improvement was due to SRL itself or the CNI reduction in the protocol when we considered the nephrotoxicity and other side effects of CNIs.25, 26 However, Toso et al.20 reported that SRL could improve survival rates in HCC patients but not in non-HCC recipients, and this suggests a specific oncological benefit of SRL. It would be interesting to determine in future studies whether SRL extends the lifespan of HCC patients by postponing death from HCC recurrence, by retarding mechanisms of aging, or by reducing CNI-related complications.
In addition, all the included studies used a combination of SRL and a low-dose CNI. As suggested by Toso et al.,20 among the various immunosuppressants, only SRL is primarily directed at HCC. Furthermore, a CNI-free protocol in which SRL is used as a primary immunosuppressive agent has been proved to be both safe and effective in kidney transplant recipients,27 and safety is guaranteed when the CNI is converted to SRL in liver transplant recipients with CNI-related renal dysfunction.28, 29 Therefore, it is important to further test whether an SRL-based, CNI-free protocol would provide maximum benefits to patients with HCC.
Increased risks of AR and HAT are 2 major concerns with SRL, although the true relationship between the use of SRL and AR or HAT is still controversial. As for the incidence of AR, no significant differences between patients on SRL-based and SRL-free regimens were found. In addition, the results of the 2 studies that were discarded from the analysis were consistent with this conclusion. This confirms that the immunosuppressive efficacy of SRL is similar to that of CNIs. Importantly, in agreement with the majority of current studies, we could not document significantly different rates of HAT in patients on SRL-based or SRL-free regimens. This result supports the reconsideration of the US Food and Drug Administration warning against SRL use in liver transplant recipients.
We also reviewed the adverse effects of SRL in this study. The most commonly reported adverse effects in the included studies were dyslipidemia, anemia, leukopenia, recurrent oral ulcers, thrombocytopenia, and peripheral edema. All adverse effects were reported to be mild and easily managed by the treatment of symptoms or the reduction of SRL.
To our knowledge, this meta-analysis is the first systematic study of all available data on SRL use in liver transplant recipients with pretransplant HCC, and it overcomes the drawbacks of each individual study (nonrandomization and small sample sizes). The 5 analyzed studies came from different countries, including the United States, China, and Italy, so the bias due to racial, diet, environmental, and etiological differences could be minimized. Overall, this study provides the most convincing results so far. However, there are some limitations. First, the patient enrollment was not randomized in all the included studies. Doctors may have a tendency to use SRL in patients with unfavorable prognostic factors for tumor recurrence or in patients suffering from concomitant conditions such as renal dysfunction due to CNIs, and this might result in a selection bias. Nevertheless, this type of bias provides even stronger support for the improved survival of patients taking SRL. Second, we were unable to perform subgroup analyses based on potential confounding factors (eg, causes of liver disease or percentages of patients meeting the Milan criteria) because of a lack of data; this may explain why we cannot draw a clear conclusion about the reason that SRL can improve the posttransplant survival of HCC patients. Third, for comparisons of the 3-year overall survival, DFS, and HAT rates, the analysis was based on only 2 studies. Under these circumstances, a single study may have had a great impact on the pooled results, and the publication bias could not be minimized. These limitations provide room for future prospective studies.
In conclusion, this study suggests that SRL is generally safe and effective and has survival and potential antitumor benefits for liver transplant recipients with a pretransplant diagnosis of HCC. It offers hope for improving the prognosis of patients with HCC and for further expanding the HCC selection criteria for OLT. Multicenter, prospective, randomized controlled studies may add more information to the findings of this study.