The incidence of hepatocellular carcinoma (HCC) is increasing in the United States and throughout the world; this trend is projected to continue for several years.1–4 Currently, HCC is the indication for approximately 750 liver transplants in the United States each year. Early experiences with transplantation in patients with HCC were poor because of a high recurrence rate. However, since the implementation of the Milan criteria, patients with HCC have been more likely to reach transplantation without tumor progression.5, 6 Nonetheless, tumor recurrence after transplantation carries an ominous prognosis; therefore, considerable effort goes into selecting candidates for whom the likelihood of recurrence is low.
Measures to prevent tumor recurrence have always been the subject of debate. Several centers have used postoperative chemotherapy to prevent tumor recurrence post-transplant,7–9 although these studies suggested that the greatest impact was in patients with advanced tumors who would generally not undergo transplantation today. Immunosuppression is a well-known risk factor for tumor growth, so it is surprising that its possible role in the outcome of liver transplantation has been poorly evaluated. Vivarelli et al.10 reported that a higher level of cyclosporine exposure was an independent prognostic determinant of tumor recurrence. Experience with sirolimus-based immunosuppression following orthotopic liver transplantation is rapidly accumulating. Sirolimus has been reported to inhibit the growth of human hepatoma cells in several in vitro studies,11 and a few clinical case reports have described the effectiveness of sirolimus in treating various malignancies in patients after liver transplantation.12–15
Despite all the in vitro evidence and sporadic case reports, it is unclear whether sirolimus has any direct oncological benefits in patients transplanted for HCC. Our aim in the current report was to examine our center's experience with sirolimus-based immunosuppression in patients who underwent liver transplantation for HCC to determine if the drug affected tumor recurrence and was safe when combined with prophylactic chemotherapy.
CI, confidence interval; HCC, hepatocellular carcinoma; HR, hazard ratio; ICU, intensive care unit; MELD, Model for End-Stage Liver Disease; NS, not significant; WBC, white blood cell.
PATIENTS AND METHODS
The study group consisted of 227 patients who underwent primary liver transplantation for HCC at the Baylor Regional Transplant Institute (Dallas, TX) between January 1995 and July 2006. Patients who exceeded the Milan criteria or had evidence of portal vein invasion by imaging were excluded. From 2000 onward, preoperative tumor ablation with chemoembolization (a selective hepatic artery infusion of 50 mg of doxorubicin, 300 mg of carbopentin, and 10 mg of mitomycin C) or radiofrequency thermal ablation was considered in all cases.
Patient data were collected prospectively and entered into an institutional transplant database. The Baylor Health Care System Institutional Review Board approved the review of these data.
One hundred six patients received tacrolimus and mycophenolate mofetil immunosuppression, and 121 patients received sirolimus-based immunosuppression. There were no consistent criteria for the selection of a specific agent over time. Although sirolimus was the preferred agent for HCC patients after 2000, many still received tacrolimus because of participation in clinical trials, clinical circumstances, or surgeon preference. Both groups received posttransplant chemotherapy as described later. The minimum follow-up for all patients was 2 years to allow an adequate assessment of tumor recurrence and a full evaluation of the safety profile of sirolimus.
On postoperative day 1, patients in the sirolimus group began receiving the drug at a dose of 2 mg orally, once daily, along with cyclosporine (Neoral, Novartis Pharmaceuticals Corp., East Hanover, NJ) at a dose of 5 mg/kg orally, twice daily. In patients with incidental tumors, sirolimus was started as soon as the explant pathology was reported, usually on postoperative day 3. Sirolimus levels were maintained at 5 to 8 ng/mL for the first 3 months and at 5 ng/mL thereafter. Cyclosporine levels were maintained at 200 ng/mL. Patients transplanted between 1995 and 2005 also received methylprednisolone; this began intraoperatively. Corticosteroids were tapered and discontinued on postoperative day 60. Since 2005, we have used a completely steroid-free protocol that includes induction with daclizumab (Zenapax, Roche Pharmaceuticals, Nutley, NJ) at 2 mg/kg on days 0 and 3 and at 1 mg/kg on day 8.
Patients in the tacrolimus group started receiving the drug at a dose of 0.05 mg/kg orally, twice daily, on day 1, and the levels were maintained at 10 to 12 ng/mL during the first 2 weeks, at 8 to 10 ng/mL during weeks 3 to 12, and at 5 to 7 ng/mL thereafter. Mycophenolate mofetil was started at a dose of 1000 mg orally, twice daily, on day 1. Patients transplanted between 1995 and 2005 received the same corticosteroids as those in the sirolimus group, and patients transplanted after 2005 followed the corresponding steroid-free protocol.
Patients with tumors discovered before transplantation were offered chemotherapy that began with 10 mg/m2 doxorubicin during the anhepatic phase of the transplant and continued postoperatively at weekly intervals for 20 weeks.7 Therapy was interrupted if patients developed leukopenia [white blood cell (WBC) count < 4000/μL] or thrombocytopenia (platelet count < 30,000/μL). Patients with HCC discovered incidentally in the explanted liver received postoperative chemotherapy if the lesion met T2 criteria and had unfavorable histological features, including vascular invasion or a poor grade, and if the patient agreed to treatment.
Liver transplantation was performed with the standard orthotopic technique.16 All patients except one received organs from deceased donors. Extra care was taken to avoid unnecessary manipulation of the tumor at the time of dissection for hepatectomy. A cell saver was not used in patients with known tumors. Liver transplantation was performed only when the tumor was confined to the liver and there was no evidence of extrahepatic spread. A careful visual examination was performed at the time of transplant to exclude any extrahepatic malignancy. If lymphadenopathy or suspicious nodules were noted in the abdomen, biopsy was performed, and frozen sections were evaluated to exclude metastatic disease before the surgeon proceeded with the transplant.
A single hepatopathologist examined all explanted livers. The number of tumors, lobar distribution of the tumors, maximal tumor diameter, histological grade according to the modified Edmondson criteria17 for the degree of tumor differentiation (grades I-IV), and presence or absence of macrovesicular and microvascular invasion were recorded. The total tumor diameter for patients with multiple tumor nodules was calculated. The new prognostic score predicting disease-free survival post–liver transplantation for HCC, recently proposed by Marelli et al.18 and Decaens et al.,19 was calculated for each patient. This score evaluates 3 variables: the largest nodule diameter (≤20 mm = 0 points, 21-30 mm = 1 point, 31-50 mm = 2 points, >50 mm = 5 points), the number of nodules (1 nodule = 0 points, 2-3 nodules = 1 point, ≥4 nodules = 2 points), and tumor differentiation (well = 0 points, moderate = 1 point, poor = 3 points). Projected 5-year survival rates were estimated with Metroticket for subjects whose tumors exceeded the Milan criteria.20
All patients were seen twice weekly in the posttransplant clinic for 3 months. Thereafter, their respective primary care physicians provided follow-up. The patients were seen monthly for the first 5 years post-transplant by the primary care physicians; laboratory tests results were faxed to Baylor University Medical Center and reviewed by an attending transplant surgeon, who performed immunosuppression changes, if there were any. Patients with abnormal clinical findings or liver dysfunction suggestive of acute cellular rejection were seen at the posttransplant clinic, and liver biopsy was performed when indicated. Rapamycin was withheld transiently if a patient had wound complications, leukopenia (WBC count < 4000/μL), or severe diarrhea. Chemotherapy was initiated after the healing of wounds and was interrupted if the WBC count was <3000/μL or if the platelet count was <30,000/μL. Patients were evaluated by the transplant surgeon and oncologist for clinically apparent recurrence or otherwise routinely at 1, 2, 5, and 10 years post-transplant, at which time a clinical examination, laboratory testing (including the serum alpha fetoprotein level), an echocardiogram, a computed tomography scan of the abdomen, and a bone scan were routinely performed to seek evidence of recurrence.
Because this was not a prospective randomized study, great care was taken to evaluate the characteristics of the patients in the sirolimus group and the tacrolimus group to identify potential bias. Using all the possible factors known in the literature to affect recurrence, we first performed a univariate analysis. The studied variables included demographic factors (age, gender, era of transplant, and waiting time prior to transplant), etiology for liver cirrhosis (hepatitis C and hepatitis B), tumor characteristics (size, number of nodules, lobar distribution, tumor grade, vascular invasion, and serum alpha fetoprotein levels), and treatment factors (use of any ablative treatment prior to transplant, use of intraoperative and postoperative chemotherapy, and use of sirolimus). Variables that were identified as potentially different by univariate analysis (P ≤ 0.10) were included in a multivariate stepwise Cox regression analysis for time to recurrence and patient survival. Recurrence-free survival and patient survival were calculated with the Kaplan-Meier method and compared with the log-rank test. Other comparisons of dichotomous variables between the groups (demographic data, complications, and tumor characteristics) were made with Fisher's exact test for 2 × 2 tables; continuous variables were compared with the Wilcoxon 2-sample test and the likelihood ratio chi-square test for larger tables, with P < 0.05 considered significant. Analysis was performed with SAS 9.1.3 (SAS Institute, Cary, NC).
The characteristics of the patients in the tacrolimus and sirolimus groups are shown in Table 1. The groups were similar with respect to age, gender, and operative characteristics. Patients in the tacrolimus group had a higher calculated Model for End-Stage Liver Disease score and were more likely to have been transplanted in the pre–Model for End-Stage Liver Disease era because sirolimus did not become available until 2000. The warm ischemia time was approximately 10 minutes longer in the tacrolimus group. More patients in the sirolimus group received ablative treatment while awaiting transplantation, as these cases occurred after 2000 when locoregional ablation was more common at our center (Table 1). However, when we separated and compared the patients who received ablative treatment (n = 99) and the patients who did not receive ablative treatment (n = 110), there was no significant difference in recurrence-free survival or patient survival (P = 0.1461). The median lengths of follow-up in the tacrolimus and sirolimus groups were 45 and 50 months, respectively.
Table 1. Characteristics of the Patients According to the Type of Immunosuppression
NOTE: Values are shown as means and standard deviations.
Abbreviation: ICU, intensive care unit; MELD, Model for End-Stage Liver Disease.
53.9 ± 7.4
53.9 ± 7.8
Number of males
Etiology of cirrhosis
Alpha fetoprotein > 200 ng/mL
Calculated MELD score
16.48 ± 7.52
13.61 ± 5.02
Era of transplant
Days on waiting list
225 ± 299
271 ± 415
Cold ischemia time (hours)
8.32 ± 2.91
7.56 ± 2.88
Warm ischemia time (hours)
0.98 ± 0.42
0.81 ± 0.25
Received preoperative ablation
Received intraoperative and/or postoperative chemotherapy
Tumor Characteristics and Tumor Prognostic Score
Characteristics of the tumors in the 2 groups are shown in Table 2. There were significantly more incidental tumors in the tacrolimus group, and this may have been related to better imaging during later years when most sirolimus patients were enrolled. The probability of recurrence of incidental and known tumors was not different (Fig. 1). There were otherwise no differences between the groups in terms of tumor size, number of nodules, proportion meeting or exceeding the Milan criteria, vascular invasion, Metroticket-estimated 5-year survival (for those exceeding the Milan criteria), or tumor grade. The tumor prognostic score18 was not statistically different between the sirolimus and tacrolimus groups (Table 3).
Table 2. Characteristics of the Tumors in the Two Groups
Tacrolimus Group (n = 106)
Sirolimus Group (n = 121)
NOTE: The columns may not add up to the same total because not all data were available for every case.
This does not include cases with >5 tumors (ie, “too numerous to count”).
NOTE: The tumor prognostic score predictive disease-free survival for this table was calculated by methods described by Marelli et al.18
The surgical complications for the 2 groups are listed in Table 4. Two patients in the sirolimus group developed arterial thrombosis versus none in the tacrolimus group, but this difference was not statistically significant. Both cases of arterial thrombosis were technical. No significant difference was found in the incidence of portal vein complications, biliary complications, wound infections, or acute rejection when the groups were compared.
Table 4. Posttransplant Complications
Tacrolimus Group (n = 106)
Sirolimus Group (n = 121)
Abbreviation: NS, not significant.
Hepatic artery stenosis
Hepatic artery thrombosis
Portal vein stenosis
Portal vein thrombosis
Bile duct stricture
Acute renal failure
New-onset diabetes mellitus
Acute cellular rejection
A higher incidence of pleural effusions occurred in the sirolimus group than in the tacrolimus group (16.53% versus 10.38%, respectively), but the difference was not statistically significant (Table 4). Sirolimus-induced pneumonitis was not seen in any of the patients.
The incidence of postoperative acute renal failure was similar for both groups. Longitudinal follow-up showed that both groups had a 40% drop in the median glomerular filtration rate at 3 months. However, the sirolimus group had a better median glomerular filtration rate at 5 years (Fig. 2).
Leukopenia (<4000/μL) was the most common side effect of sirolimus (45% of patients), and the WBC counts remained lower in this group for 3 years post-transplant. Despite this fact, there was no significant difference in the rate of infectious complications between the 2 groups. The WBC count rose in the sirolimus group after year 3, and there was no significant difference between the groups at 5 years.
Thrombocytopenia was not a problem in either group and usually normalized by 3 months after transplantation, regardless of the immunosuppressive regimen or receipt of chemotherapy. The median platelet counts (×103/μL) in the tacrolimus and sirolimus groups were 72.5 and 73, respectively, before transplant; 148 and 144, respectively, at 3 months post-transplant; 152 and 155, respectively, at 1 year; 154 and 167, respectively, at 3 years; and 176 and 175, respectively, at 5 years.
Thirteen patients (12.26%) in the tacrolimus group developed new-onset diabetes, whereas none in the sirolimus group did (P < 0.001). No significant differences were found between the groups with respect to the levels of total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, or triglycerides at 3 months, 1 year, 2 years, and 5 years. Although patients receiving sirolimus had elevated lipid levels, treatment with antihyperlipidemic agents was effective in reducing the levels to those of the tacrolimus group.
Predictors of HCC Recurrence and Patient Survival
Multivariate stepwise Cox regression analysis for HCC tumor recurrence (death censored) found that vascular invasion [hazard ratio (HR) = 6.6, 95% confidence interval (CI) = 2.65-16.44, P < 0.0001], serum alpha fetoprotein > 200 ng/mL (HR = 4.9, 95% CI = 1.77-13.56, P = 0.0005), 4 or more tumor nodules (HR = 5.3, 95% CI = 2.08-13.52, P = 0.001), and not using sirolimus (HR = 2.62, 95% CI = 0.99-6.96, P = 0.0465) were independently associated with HCC recurrence. The incidental discovery of tumors and ablative treatment prior to transplant did not influence HCC recurrence in this analysis.
Three variables were independently associated with patient death by multivariate stepwise Cox regression analysis. They included absence of sirolimus (HR = 3.0, 95% CI = 1.6-5.6, P = 0.0002), 4 or more tumor nodules (HR = 3.8, 95% CI = 1.9-7.8, P ≤ 0.0001), and presence of vascular invasion (HR = 1.8, 95% CI = 1.0-3.3, P = 0.0433).
Recurrence Probability and Patient Survival
Tumors recurred in 25 patients (11.0%) transplanted with HCC, with most occurring within the first 2 years (probability of recurrence: 5.2% at year 1, 9.2% at year 2, 10.2% at year 3, and 11.0% at year 5). The patients in the sirolimus group had a significantly lower rate of tumor recurrence (P = 0.0001) than the tacrolimus group (Fig. 3).
The sirolimus group also had significantly higher patient survival rates than the tacrolimus group at 1 year (94% versus 79%), 3 years (85% versus 66%), and 5 years (80% versus 59%, P = 0.0001; Fig. 4). When patients with incidental tumors and those with known tumors were compared, there was no significant difference in recurrence-free survival or patient survival (P = 0.3162; Fig. 5).
A significantly higher number of deaths were due to recurrent tumors in the tacrolimus group (n = 19) versus the sirolimus group (n = 4). The remaining causes of death in the tacrolimus group were recurrent hepatitis in 5 patients, sepsis in 6, de novo cancer in 6, cardiac events in 2, stroke in 2, posttransplant lymphoproliferative disorder in 2, technical reasons in 3, graft-versus-host disease in 2, graft failure in 2, other causes in 2, and unknown causes in 2. The remaining causes of death in the sirolimus group were recurrent hepatitis in 5 patients, sepsis in 5, de novo cancer in 1, posttransplant lymphoproliferative disorder in 1, technical reasons in 1, and anoxic brain injury in 2.
Sirolimus has potent in vitro effects on tumor growth.11, 21 These effects are related to the antiangiogenic properties of the drug, which are derived from the inhibition of vascular endothelial growth factor production and the stimulation of endothelial cells. Angiogenesis, or the growth of new blood vessels, occurs primarily during embryonic development and, in adults, is triggered locally in finely balanced processes such as wound healing.
Data from in vitro studies22 suggest that the effectiveness of sirolimus as an anticancer agent differs with dose, such that low nanogram per milliliter concentrations of the drug produce a potent antiangiogenic and potential antiproliferative effect, whereas higher doses may actually have a direct cytotoxic effect on tumor cells. Even though the doses used in posttransplant immunosuppression would likely be antiangiogenic, angiogenesis is only one of the mechanisms by which tumors proliferate; therefore, an antiangiogenic approach would likely be insufficient to eradicate all tumors or prevent tumor recurrence.23 Thus, combining sirolimus with other agents, including traditional systemic chemotherapy, may be logical. One could speculate that a combined approach including long-term antiangiogenic treatment with intermittent cytotoxic or other therapy might provide the best outcomes. Our posttransplant protocol for HCC patients includes doxorubicin for the first 20 postoperative weeks. Furthermore, cyclosporine is added to the immunosuppressive regimen because we have previously shown that pharmacological concentrations of cyclosporine enhance the in vitro doxorubicin cytotoxicity for multidrug resistance protein 1–positive hepatoma cells by modulating drug retention.24
Recurrence of tumors after liver transplantation for HCC is associated with a very poor prognosis: the median survival is just 13 months.25 Despite the importance of this problem, there is a paucity of clinical experience with different prophylactic regimens, including sirolimus, as strategies to reduce HCC recurrence. Kneteman et al.26 reported a pilot study of 40 patients with HCC who were treated with a sirolimus-based protocol post-transplant. The authors categorized patients within and beyond the Milan criteria, and they concluded that both the 1- and 4-year survival rates were not significantly different between the groups. They were not, however, able to address whether sirolimus affected clinical outcome. Similarly, Tosco et al.27 reported on a cohort of 70 patients with HCC who received sirolimus and showed no difference in short-term tumor-free survival in patients within or beyond the Milan criteria. Although cancer recurrence was reasonably low (11%), nearly 50% of patients suffered from delayed wound healing or incisional hernia.
Zhou et al.28 described 18 patients with HCC recurrence after liver transplantation who were switched to sirolimus-based immunosuppression from tacrolimus. Twelve of the 18 patients who had documented recurrence remained alive after 10.4 months of follow-up, and the authors concluded that sirolimus may have inhibited the progression of the recurrent tumor. However, this study had no comparison group and did not look prospectively at prevention of recurrence. Zimmerman et al.29 compared 2 groups of patients who underwent liver transplantation for HCC: 45 patients received sirolimus and a calcineurin inhibitor post-transplant, and 52 patients received cyclosporine or tacrolimus plus mycophenolate mofetil and corticosteroids. The 1- and 5-year survival rates for the sirolimus-treated group were 95.5% and 78%, respectively, versus 83% and 62% for the nonsirolimus group. The investigators found no significant difference in the incidence of major complications, and this was similar to our large experience.
Our present investigation is the first case-control study to report the usefulness of sirolimus, in conjunction with systemic chemotherapy, in preventing tumor recurrence after liver transplantation for HCC. Because of its proven cytotoxic effects on hepatoma cells in vitro, we have used doxorubicin chemotherapy for all patients undergoing liver transplantation for HCC at Baylor University Medical Center since 1987.7 To reap its clear antiangiogenic benefits, we began to add sirolimus to our tumor protocol on a limited basis in late 2000, and we combined it with doxorubicin, a cytotoxic agent capable of directly killing the residual cancer cells as they emerge from a dormant stage. We cautiously chose a lower dosage of sirolimus to avoid the potential side effects associated with a higher dosage; thus, our patients were able to receive sirolimus in the immediate posttransplant period and thereafter.
Our data suggest a beneficial effect of sirolimus immunosuppression on recurrence-free survival, which translates into patient survival benefits. At our institution's liver transplant program, we transplant patients within United Network for Organ Sharing criteria, and since December 2000, we have employed ablative treatment when feasible in patients waiting for liver transplants. One could argue that the survival benefits in the sirolimus group could be attributed to the ablative treatment that the patients received while awaiting transplantation.30 Hence, we compared the patients who received pretransplant ablative treatment with those who did not, and there was no difference in survival. We also performed a multivariate stepwise Cox regression analysis, and the ablative treatment did not influence HCC tumor recurrence. This finding suggests that the improved survival in the patients receiving sirolimus-based immunosuppression was not attributable to locoregional ablation.
The use of sirolimus during the early posttransplant period is currently challenged by the US Food and Drug Administration's black box recommendation against immunosuppression with sirolimus after liver transplantation. A recent clinical trial reported a trend toward an increased rate of graft loss and death (R. Wiesner, unpublished data, American Transplant Congress, Washington, DC, June 2003). However, sirolimus was well tolerated by our patients and did not have to be discontinued in any patient. We did not note the increase in surgical complications (eg, wound dehiscence, wound infections, and arterial thrombosis) that others have reported,27 and we suspect that our use of a lower dose of sirolimus could explain this difference. The most common side effect was leukopenia in nearly half of the patients, but it typically resolved with dose reduction. Infections were not more common in the sirolimus group. Dyslipidemia and mouth ulcers were common, as others have reported, but these were easily controlled with local or drug therapy.
The limitation of our experience is that it is a retrospective case-control study. As such, there is a potential for selection bias in this study. This is particularly important because sirolimus has been available for only the last 10 years of the study period and became our primary immunosuppressive drug of choice for tumor patients after 2000. However, the tumor characteristics in the tacrolimus and sirolimus groups were well-matched, and there were few other differences between the groups. Furthermore, multivariate analysis found that only tumor extent (vascular invasion and number of lesions) and sirolimus influenced HCC recurrence and patient survival after liver transplantation. Thus, we are confident that sirolimus was independently associated with lower HCC tumor recurrence and improved patient survival. Our data should be confirmed in a randomized prospective trial, and such a trial is currently beginning in Europe. However, this study is still in the enrollment phase, and there are no results published yet.31 Nonetheless, on the basis of our data, we can conclude that sirolimus is well tolerated at the studied doses as a treatment for patients transplanted for HCC, and it was not associated with toxicity beyond that from the systemic chemotherapy regimen. Patients who received sirolimus and posttransplant chemotherapy had better recurrence-free survival than patients who were treated with tacrolimus and mycophenolate mofetil along with posttransplant chemotherapy.