Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide, and recent epidemiologic data indicate that in western countries, the mortality rate from HCC is progressively increasing and constitutes a public health challenge.1–3
Liver transplantation (LT) is considered to be a curative strategy for patients with both HCC and cirrhosis. However, the continuing high rate of tumor recurrence represents a major drawback. In addition, most liver transplant centers are confronted with a shortage of donors and an increase in waiting time, leading to high dropout rates and HCC recurrence.4, 5 Several therapies, such as transarterial chemoembolization (TACE),6 radiofrequency ablation,7 and percutaneous ethanol injection,8, 9 have been proposed in patients awaiting liver graft to improve disease-free survival after LT. Among these different options, TACE, which has been shown to be beneficial in patient with unresectable HCC,10 has been the most widely used, but its efficiency remains controversial in this setting.11–16 Such contradictory results may be attributed, at least in part, to the wide variability in tumor spread and the characteristics of patients. Indeed, the question of whether evaluation of TACE should be restricted to patients with single or multiple nodules of HCC remains the subject of debate.
The conventional TACE technique consists of administering an emulsion of lipiodol and a cytotoxic agent via the proper hepatic artery or its right or left branch, followed by embolization. However, technical improvement is warranted because total tumor necrosis in liver explants is achieved in <20% of cases.6, 16–18 In order to more efficiently target chemotherapy to the tumor, clinicians have developed a variant of the technique called supraselective transarterial chemoembolization (STACE), which consists of catheterizing the feeding arterial hepatic branches of the tumor. In a preliminary study, we showed that STACE may lead to 50-60% total or subtotal tumor necrosis; however, we did not compare its effect with that of conventional TACE.19
The aim of this case-control study in candidates for LT for HCC was to compare the efficiency of TACE and STACE in terms of tumor necrosis in liver explants and disease-free survival. The initial analysis was performed on all patients regardless of tumor spread in order to avoid bias. In a second step, the analysis was restricted to patients with uninodular HCC ≤5 cm, considered to represent the target population for such procedures.
We reviewed all patients (n = 30) with HCC referred for LT who had undergone STACE in our liver transplant center between 1991 and 2004. Patients with incidental HCC or fibrolamellar carcinoma were excluded. These patients were matched with control patients treated with conventional TACE before LT. Control patients were independently extracted from a previous cohort of 479 subjects referred to French LT centers for HCC.20 They were selected anonymously and without knowledge of survival or HCC recurrence. Matching criteria included age, gender, Child-Pugh score, period of LT, time spent on the waiting list, size and number of tumors, and modified TNM staging system.
Charts of the 60 patients were individually reviewed, and the following data were collected and recorded: demographic information, cause and severity of liver disease, liver biochemical tests, prothrombin time, alfa-fetoprotein (AFP) level, and Karnofsky index score. Patients with cirrhosis were classified according to Child-Pugh score. Morphological features of HCC and preoperative staging were derived from abdominal imaging studies (computed tomography and/or magnetic resonance imaging). The number, size, and location of the tumors and vascular involvement were determined from this preoperative evaluation. HCCs were classified according to the Milan criteria21 and the modified TNM staging system for LT.22
TACE and STACE techniques
Conventional mesenteric arteriography was performed first, and the celiac artery was catheterized; after assessing the hepatic vascular anatomy, a mixture of 10 mL of lipiodol (Guerbet, Aulnay-sous-Bois, France) and a single anticancer drug (either 1 mg/kg of doxorubicin [Pharmacia-UpJohn, Saint-Quentin-en-Yvelines, France] or 1 mg/kg cisplatin [Lilly, St. Cloud, France]; or 1 mg/kg epirubicin [Pharmacia-UpJohn]) was provided before mechanical obstruction. For control patients, TACE was performed by nonselective injection into either the hepatic artery or its right or left branch. For STACE patients, STACE was performed by supraselective injection into the distal arteries feeding the tumor.23 Both procedures were terminated by embolization. None of the patients received pre-LT treatments other than TACE.
Post-LT management and follow-up
Data concerning time spent on the waiting list, and duration of surgery and cold ischemia, as well as pathological data on the explanted liver were obtained by reviewing the histopathological reports. Patients were screened for tumor recurrence by AFP assay and thoracic and abdominal computed tomographic scan at least every 6 months for the first 2 years, and when clinically indicated.
The χ2 test was used to compare qualitative values. Parametric (t test and the Fischer exact test) and nonparametric tests (Mann-Whitney test) were used to compare quantitative variables. Results are expressed as means ± standard deviation. Five-year disease-free survival rates were estimated by the Kaplan-Meier method and compared by the log rank test. A P value of <0.05 was considered statistically significant. Statistical analysis was performed by NCSS 2001 software.
Characteristics of Patients and HCC
Characteristics of the 60 patients are summarized in Table 1. The 2 groups were similar for all matching criteria. The only statistically significant difference was found for causes of liver disease (P = 0.008), with more alcoholic liver disease and less viral hepatitis in the STACE group than in the TACE group. There were no significant differences in pre-LT tumor characteristics (whether uninodular, number, size, modified TNM classification, or Milan criteria) (Table 2). Histological analysis of the excised tumors revealed similar tumor patterns in the 2 groups (Table 3). However, as previously reported in other studies, the number of nodules, the spread, and tumor size spread had been underestimated.
There was no significant difference in duration of surgery or cold ischemia time between the 2 groups (6.12 ± 1.6 hours and 9.7 ± 3.4 hours in the STACE group vs. 5.9 ± 2.5 hours and 7.5 ± 4.3 hours in the TACE group, respectively; P = NS). Immunosuppressive therapy after LT consisted of a dual- or triple-drug regimen including either cyclosporine A (21 patients in the STACE group and 15 patients in the TACE group) or tacrolimus (9 and 15 patients, respectively), combined with corticosteroids and/or azathioprine. The incidence of acute corticosteroid-treated rejection was similar in the STACE and TACE groups (33.3% vs. 20%, P = 0.2). Anti T cell globulin was used in 3 and 4 cases, respectively (P = 0.7). Mean follow-up after LT was 54.3 ± 41.6 months in the STACE group and 53.9 ± 35.5 months in the TACE group (NS).
Five-Year Disease-Free Survival
The median number of procedures was similar in the 2 groups (median, 1 procedure; range, 1-5 procedures; mean 1.3 ± 0.5 procedures in the STACE group and in 1.2 ± 0.7 procedures in the TACE group, P = NS). Doxorubicin was used in 85% of cases as cytotoxic agent, epirubicin in 10% of cases, and cisplatin in 5% of cases. The median waiting time for LT was 88.5 days in the STACE group and 116.5 days in the TACE group (P = NS). For all patients, disease-free 5-year survival after LT did not significantly differ between the STACE and TACE groups (76% ± 8% vs. 73% ± 8%, P = 0.6, Fig. 1). Age at transplantation (≤55 years 76% ± 8% vs. >55 years 74% ± 8%), gender (female 76% ± 14% vs. male 75% ± 6%), cause of cirrhosis (alcohol 71% ± 8% vs. virus 84% ± 7%), tumor differentiation (well differentiated 79% ± 7%, moderate 62% ± 11%, poor 66% ± 27%), tumor location (unilobar 75% ± 6% vs. bilobar 73% ± 11%), and Child-Pugh score (A 71% ± 8% vs. B/C 79% ± 8%) were not significantly associated with disease-free survival. Pre-LT AFP was significantly associated with survival in the overall population (≤12 ng/mL 93% ± 5% vs. >12 ng/mL 58% ± 9%, P = 0.002).
In a sensitivity analysis restricted to patients with a single nodule ≤5 cm (n = 32), there was a trend toward better survival in the STACE group (n = 18) than in the TACE group (n = 14) (87% ± 8.6% vs. 64.3% ± 12.8%, P = 0.09) (Fig. 2). In univariate analysis, only pre-LT AFP ≤12 ng/mL (95% ± 5% vs. 52% ± 13%, P = 0.004) was associated with disease-free survival.
In univariate analysis, the only factor associated with complete tumor necrosis was STACE in the overall population (30.8% vs. 6.9%, P = 0.02), with the same trend in the subgroup of patients with a single nodule (33.3% vs. 6.7%, P = 0.06). Total necrosis was not associated with age, gender, tumor differentiation, or cause of cirrhosis (data not shown).
Interestingly, in ideal candidates with a single nodule ≤5cm (n = 32), complete tumor necrosis seemed to be an important event, because disease-free survival was 100% in patients with total necrosis compared with 71.4 ± 9% in patients without complete necrosis (P = 0.13) (Fig. 3).
Strategies aimed at decreasing tumor recurrence are challenging for clinicians involved in the management of candidates for LT for HCC.24, 25 We observed that STACE more frequently induced complete necrosis than TACE; and that in sensitivity analysis of the ideal population for local procedures (single nodule ≤5 cm), there was a trend toward improvement of disease-free survival. Thus, attainment of total necrosis in such candidates seems to be important. We acknowledge that our study's sample size was small, which led us to draw only rough conclusions, but it provides preliminary data supporting future evaluation of STACE in a large cohort of candidates for LT with single HCC.
Pre-LT treatments are widely used; however, until now, there is no strong evidence that these treatments are beneficial or that they improve patient outcome after LT.6, 7, 9, 11–16, 20 To avoid bias, we first analyzed the impact of STACE on the total population; as expected, we did not find it to be more efficient in terms of disease-free survival. We hypothesize that this absence of evidence is mainly linked to the wide variability in tumor spread, and that locoregional ablative therapy should first be evaluated in patients with single nodules. Indeed, in a second step, we performed sensitivity analysis of patients with a single nodule ≤5 cm so we could focus on the theoretical best candidates for STACE. We found a trend toward better disease-free survival in the STACE group than in the TACE group, although case and control patients were similar in terms of known predictive factors for HCC recurrence, such as tumor size, number of nodules, bilobar nodule distribution, tumor differentiation, and macrovascular invasion. In the conventional TACE group, the 5-year disease-free survival dropped from 73% ± 8% to 64% ± 12%. We do not have a clear explanation for this finding. Our control patients were independently extracted from a previous cohort of 479 subjects referred to French LT centers for HCC. They were selected anonymously and without knowledge of survival or HCC recurrence. This paradoxical phenomenon is likely related to a random situation.
The cause of liver diseases differed greatly between the 2 groups. However, an imbalance in causes of liver disease does not seem to affect disease-free survival. In our study, cause of liver disease was not associated with disease-free survival. Moreover, in other studies with a large sample size, cause of liver disease was not identified as an independent predictive factor by multivariate analysis, in contrast to other factors such as age, number of nodules, size of nodules, and, more recently, tumor differentiation.20, 26–29 Another important finding in this study was the more complete nodule necrosis induced by the STACE procedure. This confirmed the findings of our preliminary study, which included a bias as a result of the absence of a control group.19 Moreover, patients with a single nodule ≤5 cm who achieved complete tumor necrosis had an arithmetical better disease-free survival than those without complete tumor necrosis in the STACE group compared with the TACE group. Among all pre-LT treatments, STACE might be considered a more attractive option than conventional TACE because it improves delivery of the drug to the tumor and minimizes occlusion of blood flow in the liver.23 In the present study, we clearly show that STACE was more efficient at inducing complete necrosis of HCC before LT compared with TACE. Total tumor necrosis might constitute an important objective in disease-free survival. Other competitive methods (radiofrequency ablation and percutaneous ethanol injection) might be of benefit, but half of our population had hepatocellular failure, and most patients were found to have contraindications for such procedures (low platelet count, subcapsular location).
In conclusion, this case-control study suggests that STACE performed before LT for HCC has no beneficial effect on disease-free survival in an overall population with a relatively short waiting period. However, it more frequently induces complete tumor necrosis, lending support to its usefulness in patients with a single HCC nodule. Future studies evaluating STACE in ideal candidates are warranted.