Potential conflict of interest: Nothing to report.
For the treatment of small hepatocellular carcinoma (HCC), radiofrequency ablation (RFA) is in some centers considered a first-line therapeutic option. However, such a strategy is still under debate with regard to tumor and patient characteristics. In this single-center study we assessed the 5-year survival and prognosis factors in 235 consecutive patients with cirrhosis (Child-Pugh A/B: 205/30) who received RFA as first-line treatment for up to three HCC ≤5 cm (307 tumors; mean diameter: 29 ± 10 mm; 53 multinodular forms). Among these patients, 67 satisfied the criteria for resection according to the Barcelona Clinic Liver Cancer. Complete ablation was obtained in 222 patients (94%). Overall, 337 RFA sessions were performed including iterative RFA for recurrence. Major complications occurred in three patients (0.9%), including one treatment-related death. After 27 ± 20 months of mean follow-up, local or distant, or both, tumor recurrence occurred in 16, 88, and 11 patients, respectively. Twenty-nine patients underwent transplantation and were removed from the study at this point. Overall 5-year, recurrence-free, and tumor-free (including results of iterative RFA) survival rates were, respectively, 40%, 17%, and 32%. The overall 5-year survival rate was 76% for operable patients. Factors associated with overall survival were prothrombin activity (hazard ratio [HR] = 0.97, 0.96–0.98; P < 0.0001) and serum levels of α-fetoprotein (AFP) (HR = 1.02, 1.02–1.02; P < 0.0001), and factors associated with tumor recurrence were multinodular forms (HR = 2.34; 1.52–3.6; P = 0.0001) and serum AFP levels (HR = 1.015, 1.014–1.016; P = 0.015). Tumor size was associated with local recurrence but not with overall and tumor-free survival. Conclusion: RFA is a safe and effective first-line treatment of HCC up to 5 cm in diameter, especially for patients with a single tumor, a low serum AFP level, and well-preserved liver function. (HEPATOLOGY 2009.)
In patients with cirrhosis and small hepatocellular carcinoma (HCC), radiofrequency ablation (RFA) is currently recognized as an effective local treatment.1, 2 Long-term results of RFA in a large series of patients are often difficult to compare due to the heterogeneity of selection and patient management, in light of the institution's policy for approaches used (percutaneous or intraoperative) or the indication of additional treatments, especially transarterial chemoembolization (TACE).3–6 However, there is currently a broad consensus regarding the major influence of the severity of the underlying liver disease in HCC patients.3–9 This point may in part explain discrepancies in long-term survival rates reported between Western and Eastern series. European patients are usually older, with a higher rate of alcoholic cirrhosis and a more advanced cirrhosis, explaining the fact that only a small percentage of them (<10%) are candidates for resection.10 Thus, a worse prognosis has been reported in alcoholic patients after liver resection, even when they belonged to Child-Pugh Class A.11 But in reported RFA series, the percentage of patients with cirrhosis related to alcohol is usually less than 10%.5, 12 Concerning patients with hepatitis C virus (HCV) cirrhosis, viral eradication may influence the outcome after local ablation of HCC, as suggested by case reports but, so far, poorly illustrated in large cohorts.13, 14
Furthermore, tumor parameters such as size and number of nodules or local recurrence are either contradictory or poorly evaluated.3–6, 8, 18–20 Thus, in previous series, complete ablation rates for larger HCCs (3-5 cm) after RFA seem unfavorable, ranging from 61.3% to 82.5%.21, 22 Although these results are presumably outdated thanks to recent technical progress,23 some authors suggested that RFA could be considered as an effective first-line treatment only for very small HCC (≤2 cm).12 Thus, to improve local tumor control for larger nodules, the combination of RFA with TACE is increasingly advocated.24 The impact of multinodular forms on long-term outcomes after RFA has been reported by Lencioni et al.,5 but many series include a low percentage, if any.1, 12 Data regarding the prognostic impact of iterative RFA remain scarce. As a matter of fact, iterative RFA sessions can be performed for limited recurrence more easily than resection.4, 25, 26
The purpose of this retrospective study was to evaluate the long-term results of RFA alone as first-line treatment and to determine the prognostic factors in a well-defined and closely monitored cohort of 235 consecutive Western patients with cirrhosis and early-stage HCC defined as HCC ≤5 cm diameter and fewer than three tumors.
AFP, α-fetoprotein; BCLC, Barcelona Clinic Liver Cancer; CT, computed tomography; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MRI, magnetic resonance imaging; RFA, radiofrequency ablation; TACE, trans-arterial chemoembolization; US, ultrasound.
Patients and Methods
This survey is a cohort study conducted as a retrospective analysis of a prospective database in a single center. The study was approved by the local review boards and written informed consent was obtained from all patients before treatment.
The criteria of patient selection in the present study were the following: tumor detectable by ultrasound (US), fewer than three nodules without extrahepatic metastasis, largest tumor ≤5 cm, well-compensated Child-Pugh A or B cirrhosis, prothrombin activity >40%, and platelet count >40 × 109/L. Close proximity of the colon or the hepatic hilum and abundant ascites were considered contraindications for RFA.
Between January 2001 and June 2007, 314 consecutive patients with HCC and histologically proven cirrhosis were referred to our institution for RFA. In all cases a multidisciplinary panel including a liver surgeon, an interventional radiologist, and hepatologists reviewed the patients' files and imaging examinations. Seventy-six patients were not selected for the study group because they had received, prior to RFA, treatment for HCC either by TACE (n = 12) or resection (n = 16) or because their tumors were larger than 50 mm (n = 48) (Fig. 1). Among the remaining 238 patients, three were finally not treated by RFA because of poor visualization (n = 2) or inaccessibility of the tumor (n = 1). Therefore, 235 patients with 307 HCCs (mean size: 29.2 ± 10, range 11–50 mm) were selected for this study. The baseline characteristics of the patients are reported in Table 1. In all, 182 patients had one nodule; 34 had two nodules and 19 had three nodules (all larger than 30 mm in one case). Patients with alcohol-related cirrhosis in comparison to patients with other etiologies had a significantly higher level of serum bilirubin (P = 0.006), lower prothrombin activity (P = 0.001), and a higher serum AFP level (P = 0.01). In this subgroup there was also a higher percentage of males (P = 0.002) and of Child-Pugh class B cirrhosis (P < 0.001). Among the whole population, 67 patients satisfied the criteria for resection according to the Barcelona Clinic Liver Cancer (BCLC) (Cirrhosis Child-Pugh class A with a normal level of bilirubin, no significant portal hypertension, and a single HCC). Among these, a contraindication of resection was the presence of comorbidities in five cases, age > 75 years in 18 cases, and central tumor location too close to main vessels in four cases. The remaining 40 patients were good candidates for resection but RFA was the patient's or the referring clinician's preferred choice. Nineteen patients aged less than 65 years had less than three nodules ≤3 cm or a single nodule ≤ 5 cm and Child B cirrhosis. Among these, 13 were not eligible for liver transplantation due to immunodepression related to coinfection with HIV (n = 2), contraindication for major surgery (n = 6), or persistent excessive alcohol intake (n = 5). For the other patients (n = 6), three patients declined orthotopic liver transplantation and three received RFA as bridge treatment because of the high risk of drop-out due to the estimated waiting time for a liver graft (>6 months) (n = 3).
Table 1. Characteristics of 235 Patients with 307 HCCs
HCC, hepatocellular carcinoma; SD, standard deviation; AFP, α-fetoprotein.
Patients aged >75 years, n (%)
Sex, n (%)
Etiology of cirrhosis, n (%)
Platelet count (× 109/L)
Mean ± SD
128 ± 63
Prothrombin activity (%)
Mean ± SD
75 ± 15
Serum albumin level (g/L)
Mean ± SD
40 ± 4.4
Total bilirubin level, n (%)
Serum AFP level, n (%)
Child-Pugh class, n (%)
Tumor no, n (%)
Mean ± SD (mm)
29.2 ± 10
Tumor >30 mm, n (%)
Diagnosis and Staging of HCC.
Prior to treatment, all patients underwent imaging studies including US, three-phase contrasted-enhanced computed tomography (CT) and/or magnetic resonance imaging (MRI), a physical examination, and laboratory tests. The diagnosis of HCC was histologically proven (n = 88) or established according to noninvasive criteria as defined by European Association for the Study of the Liver (n = 147).27 Liver biopsies were systematically performed when the noninvasive criteria were not satisfied.
RF Ablation Procedures.
All procedures were performed percutaneously under general anesthesia and US guidance by the same operator who had 5 years experience of percutaneous ablation at the beginning of the inclusion period. During the first 38 months, 80 patients were treated with monopolar internally cooled single or clustered electrodes (Cool Tip, Covidien/Tycohealthcare, Burlington, VT). Clustered electrodes were used systematically for tumors larger the 3 cm. Then, during the next 10 months, 78 were treated with monopolar perfused electrodes (HIIT, Berchtold/Integra, Tubingen, Germany) and finally 77 were treated with up to three internally cooled bipolar electrodes (Prosurge, Celon/Olympus, Berlin, Germany) according to tumor size. In accordance with the Society of Interventional Radiology guidelines, major complications were defined as an unexpected event increasing the level of care, and/or prolonged hospital stay and/or permanent adverse sequelae. All others were considered minor.
One month after each RFA procedure imaging examinations including US and triphasic CT or MRI, liver function tests, and measurement of serum AFP levels were performed. According to the CT or MRI results, a response to RFA was classified as complete ablation (indicated by the absence of enhancing tissue at the tumor site) or incomplete ablation (when enhancing tissue was still observed at the tumor site). In order to achieve complete ablation, the treatment course could include up to three procedures of RFA with an evaluation done 1 month after each. Therefore, persistence of active foci of tumor untreatable by RFA after a maximum of three iterative procedures was regarded as treatment failure. In this setting, other possible therapeutic options including liver transplantation, resection, TACE, or others were considered. Complete ablation achieved after a maximum of three iterative procedures spaced at 1 month were regarded as treatment success. These patients were followed by US and triphasic CT or MRI examinations and measurement of AFP serum levels every 3 months. Local recurrence was defined as a reappearance of tumor progression adjacent to the treated site and distant recurrence as the emergence of one or several tumor(s) not adjacent to the ablation zone. Whatever the type of recurrence, additional RFA treatment was considered according to the same criteria as those used at the time of the initial RFA. In other cases, patients were considered when possible for liver transplantation, resection, TACE, or others.
Categorical variables were compared with the χ2 test and continuous variables with the Mann-Whitney test; a P-value <0.05 was considered statistically significant. Overall survival was defined as the interval between treatment and death or the date of the last follow-up or the date of the most recent follow-up visit. Transplanted patients were censored from this study at the date of transplantation. Probability of recurrence-free survival was defined as the interval between treatment and the date of local and or distant HCC recurrence. Transplanted patients for whom any recurrence occurred during the waiting time of graft were censored at the date of transplantation. Tumor-free survival was defined by the absence of a detectable tumor at the endpoint date, taking into account the results of eventual iterative RFA for the treatment of recurrence. Transplanted patients for whom any active tumor was detectable on the last pretransplant imaging examinations were censored at the date of transplantation.
Univariate analysis was performed to identify clinical and biological parameters (sex, age, prothrombin activity, albumin, bilirubin levels, Child-Pugh class, serum AFP level, presence of criteria for resection [according to BCLC]), and tumor factors (size, multinodularity, and local recurrence) predicting overall, recurrence-free, and tumor-free survivals. Survival curves were computed according to the Kaplan-Meier method and compared by the log-rank test. In addition, a univariate Cox proportional hazards model was fitted to each variable. All variables with a P-value <0.05 were subjected to multivariate analysis to assess their value as independent predictors. Data analysis was performed with STATA software (release 9.0, 2006; College Station, TX).
For the treatment of initial tumors, 274 overall RFA sessions were performed: 1 in 199, 2 in 33, and 3 in 3. Complete radiological ablation was achieved in 222 patients (94.7%) (Fig. 2). Treatment failure was observed in 13/307 nodules (13 patients), including five nodules larger than 40 mm, four nodules associated with a serum AFP level >400 ng/mL, and four nodules located in areas difficult to reach. Among the 13 patients who had treatment failure after RFA, one patient was treated by resection, three patients by TACE, one patient by intra-arterial alcoholization,28 and two patients were transplanted. The other six patients also had distant multinodular recurrences. Four patients were included in clinical trials of systemic therapy and two patients with Child-Pugh B or C cirrhosis were treated by supportive care.
Survival and Recurrence
After a mean follow-up of 27 ± 20 months, three patients were lost to follow-up after 4, 6, and 13 months and were censored at this point. In all, 82 patients died, 29 were transplanted, and 121 were alive without transplantation (Fig. 2). Overall, 56 deaths were related to HCC progression, 11 to other cirrhosis complications, and 15 were unrelated to liver disease. A total of 29 patients were transplanted after a mean follow-up of 18 ± 12 months. Indications of liver transplantation were HCC recurrence in 13 patients and liver failure in 16 patients.
The estimated overall 3- and 5-year survival rates were respectively 60% and 40%, and the median of overall survival was 48 months (Fig. 3). Factors associated with overall survival are reported in Table 2. At univariate analysis prothrombin activity (P < 0.0001), albumin (P = 0.0004), total bilirubin (P = 0.0001), Child-Pugh class (P = 0.007), serum AFP level (P < 0.0001), and multinodular form (P = 0.05) were associated with overall survival. In the subgroups of 67 patients who satisfied criteria for resection according to BCLC and unresectable patients, the estimated 3- and 5-year overall survival rates were respectively 82% versus 49% and 76% versus 27% (P < 0.0001) (Fig. 4).
Table 2. Cox Survival Analysis of Predictors for Overall Survival in 235 Patients with 307 HCCs After RFA
Local tumor recurrence and tumor size either considered as continuous (P = 0.83) or categorized variables as < or >3 cm (P = 0.5) were not statistically associated with overall survival. At multivariate analysis, prothrombin activity (P < 0.0001) and serum AFP level (P < 0.0001) were found as independent predictors of overall survival.
Follow-up imaging in 27 patients (11.5%) revealed a local tumor recurrence either isolated (n = 16) or associated with distant recurrence (n = 11) (Fig. 2). In these patients the mean tumor size was 32.9 ± 10 mm (17 >30 mm). The median time to local recurrence was 19.7 months (>12 months in 19/27). Among the 27 patients with local recurrence, 16 were successfully treated by iterative RFA. Two patients were transplanted, one was treated by resection, and two by TACE. In the remaining six cases, iterative RFA was not attempted because of the presence of distant recurrences inappropriate for locoregional therapy.
Distant HCC recurrence occurred in 99 patients (42.1%), 37 were treated by iterative RFA, one by partial liver resection, nine by liver transplantation, eight by TACE, and two by intra-arterial alcoholization (Fig. 2).28 The other patients were either included in clinical trials of systemic therapy (n = 9) or treated by supportive cares (n = 33). The cumulative rate of distant recurrence at 5 years was 73%.
The estimated 3-year and 5-year recurrence-free survival rates were 37% and 18%, respectively (Fig. 5). The median of recurrence-free survival was 23 months. Factors associated with recurrence-free survival are reported in Table 3. The only factor associated with local tumor progression was tumor size. In patients with and without local recurrence, the mean tumor size was 31.6 ± 10.4 mm versus 28.3 ± 10 mm, respectively (P = 0.03). Local recurrence was not associated with tumor distant recurrence (P = 0.8). Furthermore, there was no significant difference in distant 5-year recurrence rates for patients with HCCs larger or less than 3 cm: 41% versus 38%, respectively (P = 0.94). At univariate analysis prothrombin activity (P = 0.04), serum AFP level (P < 0.02), and multinodular form (P = 0.0003) were associated with distant tumor recurrence. At multivariate analysis serum AFP level (P = 0.015) and multinodular form (P = 0.0001) were found as independent predictors of recurrence-free survival. In the subgroup of patients with HCV cirrhosis, among the 35 patients with viral C cirrhosis who received antiviral therapy, a sustained viral response was achieved in 10. For all of them, RFA achieved complete response and after a mean follow-up of 4.7 years they were all alive. With one exception, no tumor recurrence occurred in this subgroup.
Table 3. Cox Survival Analysis of Predictors of Recurrence-Free Survival in 222 Patients with 284 HCCs After Complete Ablation by RFA
At the date of the study's endpoint, among the 121 patients alive without transplantation, 101 had no detectable tumor (Fig. 2). In all, 53 (22.5%) out of the 235 patients underwent iterative RFA sessions for HCC recurrence either local (n = 12), distant (n = 37), or both (n = 4) (Fig. 2). The success rate of iterative RFA (distant and local) was 87%.
Incorporating results of iterative RFA treatments for HCC recurrences the estimated 3- and 5-year rates of tumor-free survival were 56% and 32%, respectively (Fig. 6). The median of tumor-free survival was 38 months. Factors associated with tumor-free survival are reported in Table 4. At univariate analysis prothrombin activity (P = 0.009), serum AFP level (P = 0.015), and multinodular form (P = 0.002) were associated with tumor-free survival. At multivariate analysis prothrombin activity (P = 0.041), multinodular form (P = 0.006), and serum AFP level (P = 0.008) were found as independent predictors of tumor-free survival.
Table 4. Cox Survival Analysis of Predictors of Tumor-Free Survival* in 222 Patients with 284 HCCs After Complete Ablation by RFA
Including results of repeated RFA for eventual recurrence.
Age (per 1 year)
Platelet count (×109/L)
Prothrombin activity (%)
Serum albumin (g/L)
Total bilirubin (mg/dL)
Child-Pugh class (B)
Tumor number (>1)
Tumor size (mm)
Complications of RFA
Overall, 337 RFA sessions were performed including 63 RFA sessions for the treatment of recurrence. Major complications occurred in three patients (0.9%): one death related to unexplained shock occurred 2 days after the procedure. Another patient developed a neoplastic seeding of the needle track after treatment for a large subdiaphragmatic distant recurrence 2 years after the first RFA of initial HCC. The patient was treated by costal resection but developed metastasis in the lung and died 2 years later. The third complication was a liver abscess related to colonic perforation, which occurred after RF ablation of a tumor abutting the colon in a patient with antecedents of multiple abdominal surgeries. The abscess was successfully treated by drainage and adequate antibiotics.
As previously reported,3–9 we observed that liver function strongly influenced overall and HCC-free survival rates. Thus, prothrombin activity, albumin, total bilirubin, and Child-Pugh class were among the strongest predictors of outcome with the serum AFP level and multinodularity. Liver function was a particularly important factor in our cohort, a fact that was in part explained by the high percentage of patients with alcoholic cirrhosis. Liver function of these patients was significantly worse in comparison to that of patients with cirrhosis of other etiologies. The overall survival rate observed in this study (40% at 5 years) was lower than that reported in other studies, even Europeans mainly from Italy and including patients with viral C cirrhosis who usually have less affected liver function. However, by selecting patients who satisfy BCLC criteria for resection, the 5-year overall survival rate reached 76%, which is comparably even better than that reported in other series, including for patients with similar or better baseline characteristics and who were treated either by RFA or resection.1, 12, 29, 30
In our study, in addition to high serum AFP and multinodular forms, low prothrombin activity was also a predictive factor of distant HCC recurrence. This suggests that the severity of the underlying liver disease, which is a risk factor for HCC occurrence, may also be a risk of HCC recurrence, reinforcing the importance and role of the liver status in hepatocarcinogenesis. Interestingly, in patients with HCV cirrhosis we also observed that sustained viral response to antiviral treatment was associated with a drastic decrease of HCC recurrence and a high survival rate. Not surprisingly, a high serum AFP level was predictive of HCC recurrence and poor prognosis.31 A high serum AFP level is usually observed in tumors of high-grade malignancy. Furthermore, a moderate increase of AFP, unrelated to tumors, is a well-known risk factor of HCC occurrence in cirrhotic livers.32, 33
An important observation of our study is the fact that despite the tumor's size as a predictor of local recurrence, this parameter, whether considered as a continuous or discontinuous variable, and local recurrence itself, did not impact overall and tumor-free survival. Interestingly, similar results have been recently reported by Lam et al.4 The main explanations of these apparently paradoxical results are, first, that the local recurrence rate observed (11.5%) was quite low in comparison with results reported by earlier studies including medium-sized (>3-5 cm) HCCs,22, 34 presumably because RFA has improved technically35 and technologically23, 36 over time; and second, that the majority of these local recurrences (59%) was sufficiently limited to be completely ablated by iterative RFA, highlighting the benefit of careful posttreatment follow-up to allow additional procedure as soon as possible.
Converse to liver resection, which can only be repeated in a few distant recurrence cases,25, 26 our study clearly shows that iterative RFA is also effective to achieving complete and durable ablation in the majority of cases. Thus, in this survey among the 46% (53/116) of patients with recurrences, of whatever type, iterative RFA achieved complete response for 87%.46/53 Therefore, the results of RFA should not only be considered in terms of recurrence-free survival, but also in terms of tumor-free survival. In this study the median recurrence-free survival was 23 months, but when the results of iterative RFA were taken into account, the median tumor-free survival was 38 months.
Because, in our experience, iterative RFA remains a safe and very effective treatment in the majority of cases of limited recurrences, the need to resort to “salvage” surgical treatments either by resection (n = 3) or liver transplantation (n = 13) was quite low. As for TACE, the indication was limited to only 5.5% (n = 13) of patients with multinodular relapses unsuited to any local therapies or liver transplantation and who had sufficiently preserved liver function. The remaining patients with advanced stages of recurrence (n = 58) received supportive care or were enrolled into pharmaceutical trials of drugs, which have a marginal impact on survival.37, 38 However, the fact that a nonnegligible ratio of our patients (12.3%, 29/235) finally underwent liver transplantation may have favorably influenced survival figures of the study. However, this possibility seems slight because all transplanted patients were censored at the date of transplantation and only 13 of them had detectable viable HCC at this time. Moreover, in this study the quite long waiting times after RFA for transplantation (mean: 18 ± 12 months), support the effectiveness of RFA as a bridging treatment for transplantation.39
Only three (0.9%) severe complications occurred, confirming the safety of RFA. Among these, one needle track seeding (1/337 procedures) was observed. Thus, the rate of this complication was within the range of those usually reported.40
In conclusion, RFA appears as a safe and effective treatment in patients with early-stage HCCs up to 5 cm in diameter. Within this limit, the size of HCC is a predictor of local recurrence but not of overall or tumor-free survival because most of these recurrences are limited and therefore easily entirely ablated by additional RFA sessions. Despite a complete ablation rate higher than 90%, patients' overall, recurrence-free, and tumor-free survival rates were significantly impacted by liver function, high basal serum AFP level, and the presence of multinodular forms. However, compared with the results of partial liver resection, survival rates after RFA are similar in operable patients with Child-Pugh A, single tumors, normal bilirubin level, and no significant portal hypertension. Therefore, resorting to RFA for early HCC as a first-line treatment even for patients usually considered good candidates for surgery appeared justified, especially regarding its drastically lower rate of complication and its higher repeatability in the case of recurrence. Furthermore, for patients eligible for liver transplantation who had basically bad prognostic factors, such as liver dysfunction, a high serum AFP level, and more than one nodule, RFA seems capable of providing an effective bridge to transplantation.