Percutaneous radiofrequency ablation for hepatocellular carcinoma

An analysis of 1000 cases




Radiofrequency ablation (RFA) was introduced recently as a therapeutic modality for hepatocellular carcinoma (HCC), an alternative to percutaneous ethanol injection therapy (PEIT), which is coming into use worldwide. Previously reported treatment efficacy and complication rates have varied considerably.


Between February 1999 and February 2003, the authors performed 1000 treatments of RFA to 2140 HCC nodules in 664 patients with a cooled-tip electrode at the University of Tokyo Hospital (Tokyo, Japan). Short-term and long-term complications were analyzed by treatment and session basis. Cumulative survival was also assessed in 319 patients who received RFA as primary treatment (naive patients) and 345 patients who received RFA for recurrent tumor after previous treatment including resection, PEIT, microwave coagulation therapy, and transarterial embolization (nonnaive patients).


A total of 40 major complications (4.0% per treatment, 1.9% per session) and 17 minor complications (1.7% per treatment, 0.82% per session) were observed during the observation period until March 31, 2004. There were no treatment-related deaths. Surgical intervention was required in one case each of bile peritonitis and duodenal perforation. The cumulative survival rates at 1, 2, 3, 4, and 5 years were 94.7%, 86.1%, 77.7%, 67.4%, and 54.3% for naive patients, whereas the cumulative survival rates were 91.8%, 75.6%, 62.4%, 53.7%, and 38.2% for nonnaive patients, respectively.


The authors confirmed the safety and efficacy of RFA for HCC in a large-scale series and long-term prognosis was satisfactory. Cancer 2005. © 2005 American Cancer Society.

Hepatocellular carcinoma (HCC) is a common malignancy worldwide, with an increasing incidence in the United States.1, 2 Current options for the treatment of this cancer consist of surgical resection, transcatheter arterial embolization (TAE), and percutaneous ablation therapy. Although surgical resection usually is considered to be the first-choice treatment,3, 4 it is not infrequently contraindicated by underlying chronic liver diseases based on hepatitis B or C virus (HCV) infection.5, 6 Orthotopic liver transplantation (OLT) is a strategy that can treat both cancer and liver dysfunction, and indeed has shown excellent survival in patients at an early stage of the cancer (e.g., a single nodule of ≤ 5 cm in dimension or < 3 nodules of ≤ 3 cm in dimension).7, 8 However, with an increasing demand for donor tissue but a limited supply, the waiting time for an OLT is now > 1 year in Europe and the United States.8, 9 TAE is a widely performed procedure for patients with multiple hypervascular nodules. However, complete necrosis rarely is achieved.10 Percutaneous ethanol injection therapy (PEIT) continues to play an important role in the treatment of small HCCs, and is considered to be a bridge treatment for OLT. Long-term outcomes after PEIT, such as 5-year survival, are comparable to those of surgery.11, 12 However, the efficacy of PEIT is dependent on tumor size, and a local tumor progression rate of > 20% was reported for tumors > 3 cm.13, 14

Radiofrequency ablation (RFA) is a recently introduced alternative technique to PEIT and is rapidly gaining use worldwide.15–19 An area of ≤ 3 cm in diameter can be ablated with a single application of RFA.20, 21 The predictability of the ablation area is one of the merits of RFA compared with PEIT. The treatment efficacy and complication rate of RFA have been described in various studies.22, 23 We have performed 1000 treatments of RFA to 2140 HCC nodules in 664 patients and we assessed the early and late complication rates of RFA as well as long-term outcomes in this large case series.



Since we introduced RFA to our department in February 1999, we had treated 664 patients with HCC by RFA in 4 years by February 28, 2003. All those patients were included in the current study. Ablation therapy was used either because patients were considered not to be suitable for resection (n = 419) by consideration of liver function impairment, number and distribution of the tumors, and cardiopulmonary dysfunction, or because they voluntarily preferred ablation with informed consent (n = 245) despite surgery also being feasible. Inclusion criteria for RFA were as follows: total bilirubin concentration < 3 mg/dL, platelet count ≥ 5 × 105/mm3, and prothrombin activity ≥ 50%. Ascites should be controlled beforehand by diuretics. Patients with portal vein tumor thrombosis or extrahepatic metastasis were excluded. We also excluded patients who had a history of bilioenteric anastomosis or sphincterotomy that are considered as high risk for hepatic abscess formation. Written informed consent was obtained from all enrolled patients, and the protocol was approved by the ethics committee of the University of Tokyo Hospital (Tokyo, Japan).

Technical Terms

We defined a session as a single intervention episode that consists of one or more ablations performed on one or more tumors and a treatment as the completed effort to ablate one or more tumors that consists of one or more sessions according to the working party report on image-guided tumor ablation.24

Diagnosis of Hepatocellular Carcinoma

HCC was diagnosed based on typical findings by ultrasonography and computed tomography (CT) scans (hyperattenuation in the arterial phase and hypoattenuation in the portal-venous phase). The diagnosis of HCC also was confirmed histopathologically with ultrasound-guided biopsy in 302 of 319 naive patients. Ultrasonography was performed with a 3.5–6.0-MHz convex probe using an Aloka SSD-5500 (Aloka, Tokyo, Japan) or Toshiba SSA-370A machine (Toshiba Medical Systems, Tokyo, Japan). The dimension of the tumor nodule was measured by visualization of the largest dimension of the tumor.

Using dual-phase dynamic CT scans, arterial and portal phase images were obtained 31–33 seconds and 120 seconds, respectively, after the initiation of contrast material injection. Spiral CT scans were performed with 5-mm–thick sections at a table feed speed of 5–7 mm/sec. Using CT scans during arterial portography, scanning began 30 seconds after contrast medium injection. Using CT scans during hepatic arteriography, images were obtained after injection of 30 mL of contrast medium at a rate of 1.2 mL/sec. When > 3 hypervascular nodules were detected or the largest nodule was > 3 cm on these examinations, we subsequently performed TAE. Immediately after injection of 2–10 mL of iodized oil (Lipiodol Ultra-Fluid, Shering Japan, Osaka, Japan), feeding arteries were embolized selectively with gelatin sponge particles (Spongel, Yamanouchi Pharmaceutical Co., Tokyo, Japan). Otherwise, 1–2 mL of iodized oil only was injected to intensify the radiologic visibility of the target nodules. In February 2003, we changed the size criterion for performing TAE from > 3 cm to > 2 cm in diameter.

Specimens for histologic evaluation were obtained by ultrasound-guided needle biopsy using a 20-gauge needle (Bard Monopty, C.R. Bard, Inc., Covington, GA). Histopathologic grading of tumor differentiation was done according to the criteria of Edmondson and Steiner.25

Electrode Insertion and Radiofrequency Ablation

A 17-gauge, cooled-tip electrode was inserted under real-time ultrasound guidance. For the intercostal approach, we adjusted the operating table so that patients had a rotated or head-up position to allow for electrode insertion as square to the thoracic wall as possible. For the subcostal approach, patients usually remain in the head-up position to allow the electrode to be inserted without the patient taking a deep breath. We also utilized an intrapleural infusion technique when the tumor was located adjacent to the diaphragm to visualize the entire image of the tumor.26 A glucose solution (5%, 500 mL) was infused into the right pleural cavity under ultrasonographic guidance before needle insertion.

An electrode with a 2-cm or 3-cm exposed tip was connected to a 500-kHz RF generator (Radionics, Burlington, MA), which produces 200 W at 50Ω of impedance.20, 27 The equipment also allows the measurements of generator output, tissue impedance, and electrode tip temperature. A tip temperature of 10–20 °C is maintained by a peristaltic pump infusing chilled saline solution. After insertion of the electrode into the lesion, we started ablation at 60 W for the 3-cm exposed tip and 40 W for the 2-cm exposed tip. The power was increased to 140 W at a rate of 20 W/min. When a rapid increase in impedance was observed during thermal ablation, we minimized the output for 15 seconds and restarted the emission at a lower output.28 The duration of a single ablation was 12 minutes for the 3-cm electrode and 6 minutes for the 2-cm electrode. After RF exposure, the pump was stopped and the temperature of the needle tip was measured. When the temperature was < 65 °C, additional ablation was performed. When the target nodule was > 2 cm in diameter, we performed multiple ablations. When the total ablation time in a treatment was > 60 minutes, we divided a treatment into ≥ 2 sessions in consideration of burden to the patient.

Antibiotics were administered before and after the procedure on the treatment day and on the morning of the next day. We continued to administer antibiotics when the patients had a fever > 37.5 °C.

Assessment of Treatment Efficacy and Follow-Up

After 1–2 sessions of RFA, dynamic CT scans with section thickness of 5 mm were performed to evaluate the ablation. Complete ablation of HCC was defined as hypoattenuation of the lesion including the surrounding liver parenchyma. Patients received additional sessions of RFA until the treatment was judged as complete. Follow-up consisted of monthly blood tests and monitoring of tumor markers at the outpatient clinic, and ultrasonography and dynamic CT scans were performed every 3–4 months. Intrahepatic HCC recurrence was classified as either tumor recurrence at a site distant from the primary tumor or adjacent to the treated site (local tumor progression). If HCC recurrence was suspected, further examinations including dynamic CT scans and CT angiography were performed. When recurring HCC tumors were identified, RFA was performed if the same initial inclusion criteria were again satisfied.

Complications were assessed on the basis of the number of treatments and sessions. Major complications were defined as those that, if left untreated, might threaten the patient's life, lead to substantial morbidity and disability, or result in hospital admission or substantially lengthen hospital stay according to the previously described guidelines.24 All other complications were considered minor. Complications were classified into 3 categories according to the time after the last ablation: immediate complications (≤ 6–24 hours after the procedure), periprocedual complications (within 30 days), and delayed complications. In assessing delayed complications, follow-up was censored either on March 31, 2004 or in the event of death.

Survival analysis was performed on patient basis. Survival time was defined as the interval between the first RFA and the death or the last visit to outpatient clinic until March 31, 2004. Six-hundred sixty-four patients were divided into 2 groups: 319 patients who received RFA as the initial treatment for HCC (naive patients) and 345 patients who received RFA as a treatment for tumor recurrence after primary treatments including hepatic resection, PEIT, percutaneous microwave coagulation therapy (PMCT), and TAE. A cumulative survival curve was plotted using the Kaplan–Meier method. Survival was also assessed in naive patients among subgroups divided by age, gender, etiology of background liver disease, Child–Pugh class, tumor size and number, pathologic grade of tumor, and tumor markers. Difference between these subgroups was tested by the log-rank test.

Tumor recurrence was analyzed among 306 of the 319 naive patients, excluding 13 in whom several nodules were left untreated with the initial RFA therapy. The observation period for tumor recurrence was defined as the interval between the first RFA and either of the detection of tumor recurrence, death, or the last visit before March 31, 2004, whichever came first. As death and tumor recurrence are competing risks, we used cumulative incidence estimation with competing risks for the analysis of cumulative tumor recurrence as described by Gray.29 Differences with P < 0.05 were statistically significant. All statistical analyses were performed with S-plus 2000 software (MathSoft, Inc., Seattle, WA).


Patient Profile

The mean age of the patients was 67 years (range, 44–90 years). Patients were male dominant (69%) and the majority were HCV positive (Table 1). The mean tumor size was 2.6 cm (range, 0.8–9.7 cm). As 227 of 664 patients received ≥ 2 treatments during the period for tumor recurrence, the total number of treatments (cases) received was 1000.

Table 1. Characteristics of 664 Patients Treated by RFA
VariablesNaive patients (n = 319) (%)Nonnaïve patients (n = 345) (%)
  • RFA: radiofrequency ablation; HBs-Ag: hepatitis B surface antigen; HCV-Ab: hepatitis C virus antibody; AFP: α-fetoprotein; DCP: des-gamma carboxyprothrombin.

  • a

    Values are expressed as mean ± standard deviations.

Mean age (yrs)a67.4 ± 7.866.6 ± 8.4
Male gender212 (66.5)247 (71.6)
 HBs-Ag positive only31 (9.7)36 (10.4)
 HCV-Ab positive only251 (78.7)284 (82.3)
 Both positive5 (1.6)6 (1.9)
 Both negative32 (10.0)19 (5.5)
Alcohol consumption >80 g/day52 (16.3)59 (17.1)
Child–Pugh class  
 A221 (69.3)225 (65.2)
 B94 (29.5)111 (32.1)
 C4 (1.3)9 (2.6)
Tumor size (cm)  
 ≤2.087 (27.2)113 (32.8)
 2.1–5.0215 (67.4)209 (60.6)
 >5.017 (5.3)23 (6.7)
Tumor no.  
 Single193 (60.5)138 (40.0)
 2–3105 (32.9)141 (40.9)
 >321 (6.6)66 (19.1)
Tumor markers  
 AFP (ng/mL)  
 ≤20152 (47.6)156 (45.2)
 21–10086 (27.0)97 (28.1)
 101–40047 (14.7)43 (12.5)
 >40034 (10.7)49 (14.2)
 DCP (mAU/mL)  
 ≤40225 (70.5)226 (65.5)
 41–10036 (12.9)49 (14.2)
 101–40037 (11.6)40 (11.6)
 >40021 (6.6)30 (8.7)

Treatment Efficacy and Complications

One thousand treatments to 664 patients consisted of 2082 sessions of RFA to 2140 nodules. The distribution of nodules was as follows: 367 nodules (17.1%) in the lateral segment, 238 (11.1%) in the medial segment, 891 (41.6%) in the anterior segment, 609 (28.5%) in the posterior segment, and 35 (1.6%) in the caudate lobe. The mean number of electrode insertions according to tumor size are as follows: 1.5 times for nodules < 2 cm, 2.3 times for nodules 2.1–3.0 cm, 4.2 times for nodules 3.1–5.0 cm, and 11.7 times for nodules > 5 cm. One hundred thirty-seven of 319 naive patients and 145 of 345 nonnaive patients received TAE before RFA. All nodules were ablated completely in 934 treatments. RFA was performed for debulking the tumor burden in the remaining 66 treatments. The α-fetoprotein (AFP) declined to < 100 ng/mL in 184 of 247 treatments (74.4%) in which the pretreatment AFP level was > 100 ng/mL. DCP normalized (≤ 40 mAU/mL) in 279 of 326 treatments (85.6%).

A total of 40 major complications (4.0% per treatment, 1.9% per session) and 17 minor complications (1.7% per treatment, 0.82% per session) were observed within the follow-up period (Tables 2, 3). There were no complication-related deaths. Side effects such as moderate pain controlled by analgesics, nausea, or infection-unrelated fever relieved by antifebriles were not included. As for major complications, intraperitoneal hemorrhage was observed in four patients. In one patient, bleeding was detected during RF exposure, and arterial hemorrhage was identified by color Doppler ultrasonography. We reablated the needle track to stop the bleeding. The remaining three patients recovered after transfusion under careful observation. Hepatic infarction, defined as an increase in serum aspartate aminotransferase levels > 1000 IU/mL and/or an appearance of a wedge-shaped hypoattenuated area no smaller than a subsegment by dynamic CT scan, occurred in 2 patients. Both recovered spontaneously, although they suffered from high-grade fever lasting for 1–2 weeks. Hepatic abscess formation was found in seven patients and percutaneous drainage was performed in all. Two of them were complicated with bronchobiliary fistula and recovered after endoscopic biliary drainage. Intestinal perforation/penetration (duodenum, stomach, or colon) was observed after ablation in three patients. One patient with duodenal perforation was complicated with peritonitis and treated by surgical intervention. Colonic penetration occurred in one patient after hepatic abscess. The abscess was treated by percutaneous drainage and the anastomosis to the colon was closed by colonoscopic procedure. The patient with gastric penetration recovered with total parenteral nutrition without surgical intervention. In 1 patient, bile peritonitis occurred approximately 12 hours after RFA, requiring surgical and endoscopic biliary drainage. The patient showed intrahepatic bile duct dilatation before RFA, probably due to bile duct damage by previous PMCT.

Table 2. Major Complications after RFA
ComplicationsNo. of complicationsPrevalence (%)
Per treatmentPer session
  1. RFA: radiofrequency ablation.

Immediate (within 24 hrs)   
 Intraperitoneal hemorrhage requiring blood transfusion40.40.19
 Pleural effusion requiring drainage40.40.19
 Hepatic infarction20.20.096
 Pneumothorax requiring drainage10.10.048
 Hemothorax requiring drainage10.10.048
 Bile peritonitis10.10.048
Periprocedual (within 30 days)   
 Hepatic abscess requiring drainage70.70.34
 Bronchobiliary fistula20.20.096
 Duodenal perforation10.10.048
 Colonic penetration10.10.048
 Gastric penetration10.10.048
 Neoplastic seeding151.50.72
Table 3. Minor Complications after RFA
ComplicationsNo. of complicationsPrevalence (%)
Per treatmentPer session
  1. RFA: radiofrequency ablation.

Immediate (within 24 hrs)   
 Self-limiting hemobilia30.30.14
 Skin burn (entrance point)30.30.14
Periprocedual (within 30 days)   
 Self-limiting portal vein thrombosis40.40.19

When the 1000 treatments were divided into 2 groups as the earlier 500 and the later 500, there was a trend toward a decrease in the rates of immediate and periprocedual major complications (from 3.4% to 1.6% per treatment, P = 0.105 by chi-square test).

As a late complication, carcinoma seeding was identified in 15 patients with a median occult period of 17 months after the last ablation (range, 2–28 months). The area of tumor seeding was localized in nine patients and was surgically resected. Microwave coagulation was performed under thoracoscopic guidance in one patient with dissemination to the pleura. One patient with localized spread of tumor received radiotherapy. Two patients with intraperitoneal tumor spread were treated by systemic chemotherapy. The remaining two patients received best supportive care, as their liver function was impaired severely.

Long-Term Outcomes

Of the 664 patients included in the current series, 203 had died by March 31, 2004. Eight patients (1.2%) were lost to follow-up. The median observation period was 2.3 years (range, 0.17–5.1 years). Causes of death were cancer progression in 136 patients, hepatic failure in 36 patients, upper gastrointestinal bleeding in 8 patients, and liver-unrelated causes in 23 patients. The cumulative survival rates estimated by the Kaplan–Meier method at 1, 2, 3, 4, and 5 years were 94.7%, 86.1%, 77.7%, 67.4%, and 54.3% for naive patients whereas the respective rates were 91.8%, 75.6%, 62.4%, 53.7%, and 38.2% for nonnaive patients, respectively (Fig. 1). Significant differences were observed in subgroups divided by Child–Pugh class (P = 0.000388), tumor size (P = 0.000234), AFP level (P = 0.0103), and DCP (P = 0.00000473; Fig. 2 and Table 4). There was no significant difference in subgroups divided by age (P = 0.378), gender (P = 0.921), etiology of background liver disease (P = 0.909), number of tumors (P = 0.464) or pathologic grade (P = 0.111).

Figure 1.

The cumulative survival rates of naive (thin line) and nonnaive (thick line) patients who received radiofrequency ablation (RFA). The cumulative survival rates estimated by the Kaplan–Meier method at 1, 2, 3, 4, and 5 years were 94.7%, 86.1%, 77.7%, 67.4%, and 54.3% for naive patients and 91.8%, 75.6%, 62.4%, 53.7%, and 38.2% for nonnaive patients, respectively.

Figure 2.

Cumulative survival rates of naive patients divided by (A) Child–Pugh Class and (B) tumor size. (A) The cumulative survival rates at 1, 2, 3, 4, and 5 years were 96.4%, 90.4%, 83.4%, 72.9%, and 63.1% for Child–Pugh A patients (thin line) and 90.7%, 79.0%, 65.0%, 53.9%, and 31.4% for Child–Pugh B/C patients (thick line), respectively. (B) The cumulative survival rates at 1, 2, 3, 4, and 5 years were 100%, 93.2%, 90.8%, 90.8%, and 83.8% for patients with tumors ≤ 2 cm (thin line), 93.0%, 85.4%, 74.3%, 63.0%, and 45.2% for patients with tumors 2.1–5.0 cm (thick line), and 87.5%, 73.4%, 58.7%, 33.6%, and 33.6% for patients with tumors > 5.0 cm (dashed line), respectively. RFA: radiofrequency ablation.

Table 4. Three-Year Survival of Patients with HCC Treated by RFA
ParameterNo. of patients3-Year survival rate (95% CI)P value
  1. HCC: hepatocellular carcinoma; RFA: radiofrequency ablation; HBsAg: hepatitis B surface antigen; HCV-Ab: hepatitis C virus antibody; AFP: α-fetoprotein; DCP: des-gamma carboxyprothrombin.

Overall31977.7 (72.4–83.2) 
Age (yrs)   
 >6815976.0 (68.5–84.2)0.378
 ≤6816079.2 (72.1–87.1) 
Gender  0.921
 Male21277.7 (71.5–84.5) 
 Female12777.1 (67.8–87.8) 
Etiology  0.909
 HBs-Ag positive only3080.5 (66.3–97.8) 
 HCV-Ab positive only25278.3 (72.5–84.6) 
 Both negative3370.8 (54.1–92.6) 
Child–Pugh class  0.000388
 A22183.4 (77.7–89.5) 
 B/C9865.0 (54.9–76.3) 
Tumor size (cm)  0.000234
 ≤2.08790.8 (83.8–98.4) 
 2.1–5.021574.3 (67.7–81.5) 
 >5.11758.7 (38.0–90.8) 
Tumor no.  0.464
 Single19377.3 (70.1–85.2) 
 2–310577.5 (69.2–86.9) 
 >32167.5 (48.6–93.6) 
Edmondson grade  0.111
 16183.7 (72.6–96.7) 
 221476.9 (70.5–93.9) 
 32766.0 (48.9–89.1) 
AFP (ng/mL)  0.0103
 ≤10023882.3 (76.9–88.1) 
 101–4004772.2 (58.6–88.8) 
 >4003447.6 (28.6–79.4) 
DCP (mAU/mL)  0.00000473
 ≤4022586.3 (81.4–91.5) 
 41–1003662.4 (46.6–83.4) 
 >1005856.1 (42.4–74.4) 

Tumor recurrence was identified in 165 of 306 patients. The median observation period for tumor recurrence was 1.59 years (range, 0.12–4.98 years). Of 165 patients with tumor recurrence, 142 (86.1%) had ≤ 3 recurrent nodules. Among these 142 patients, 137 were treated by second RFA, 4 by ethanol injection, and 1 by hepatic resection. Most of the remaining 23 patients with multiple recurrent nodules (> 3) were treated by TAE. Cumulative probabilities of tumor recurrence at 1, 2, 3, and 4 years were 20.4%, 43.4%, 59.8%, and 65.9%, respectively. Cumulative probabilities of death without tumor recurrence at 1, 2, 3, and 4 years were 3.9%, 7.0%, 8.7%, and 10.9%, respectively (Fig. 3A). When the patients were divided by Child–Pugh class, there was no significant difference in overall recurrence rate between Class A and Class B/C patients (P = 0.38 by Gray's test), whereas a significant difference in death without tumor recurrence was observed between the 2 groups (P = 0.0026; Fig. 3B). Cumulative probabilities of local tumor progression at 1, 2, 3, and 4 years were 1.3%, 2.4%, 2.4%, and 2.4%, respectively. (Fig. 4). No extrahepatic metastasis was observed in these 306 patients except for 2 patients with neoplastic seeding.

Figure 3.

(A) Cumulative probabilities of tumor recurrence (thin line) and death without tumor recurrence (dotted line) of patients who received radiofrequency ablation (RFA) as initial treatment for hepatocellular carcinoma (HCC). The sum of the 2 probabilities is equivalent to 1 − tumor recurrence-free survival (thick line). (B) Cumulative probabilities of tumor recurrence and death without tumor recurrence of patients who received RFA as initial treatment for HCC divided by Child–Pugh class. There were significant differences between the cumulative probabilities of death without tumor recurrence of Child–Pugh A and Child–Pugh B/C patients, whereas no significant difference was observed between the cumulative tumor recurrence probabilities.

Figure 4.

Cumulative probabilities of overall tumor recurrence (solid line) and local tumor progression (dotted line). RFA: radiofrequency ablation.


In the current study, we have shown that RFA is a safe and effective strategy for the treatment of HCC. The incidence of major complications within 30 days after the procedure was satisfactorily low (2.5% per treatment), and there were no treatment-related deaths.

As an immediate complication, intraperitoneal bleeding is a potentially fatal condition. Thus, a careful estimation of the underlying coagulopathy is important to avoid hemorrhage. We set the minimum of prothrombin time and platelet count for the indication of RFA at higher levels (> 50% and > 5.0 × 104/mm3) than those for PEIT.12 The observed frequency of peritoneal bleeding (0.4% per treatment) was similar to that previously reported (0.3–0.5%).22, 23, 30 Gastrointestinal perforation, another potentially fatal complication, can occur when the target nodule is located adjacent to the intestine, especially when the patients have a previous history of gastrointestinal surgery.22, 23 The observed frequency of gastrointestinal perforation/penetration (0.3% per treatment) was the same as previously reported, although none of those patients had a history of surgery. We recently introduced an intraperitoneal infusion technique, by which 500–1000 mL of 5% glucose solution is injected before and during the ablation for the purpose of creating space between the lesion and the intestine. The efficacy of this procedure is yet to be assessed fully.

It is well understood that the risk of complications can be reduced by proficiency in technique and refinement in pretreatment assessments. We observed a trend toward a decrease in the rates of immediate and periprocedual complications (from 3.4% to 1.6% per treatment). Conversely, we encountered an increase in the number of patients with neoplastic seeding. The main reason for that is the prolonged follow-up period because the majority of those patients were identified > 17 months after the last ablation. The rate of neoplastic seeding (1.5%) was higher than that previously reported (0.2–0.5%), although it was still very low compared with the value (12.5%) reported by Llovet et al.31 The high incidence of seeding was likely to be related directly to the finding that tract cauterization was not performed.23 Another possible reason is the tumor biopsy we performed in most of the patients to obtain the pathologic confirmation.32

The overall tumor recurrence rate for RFA in the current study was compatible with that of PEIT and hepatectomy.33–35 The local tumor progression rate was 2.4% during a median of 19 months of follow-up, a very low rate compared with that reported elsewhere. Local tumor progression after RFA occurs mainly at the surrounding tissue.36, 37 To completely ablate a tumor ≥ 2 cm in diameter with a sufficient safety margin, multiple electrode insertion is necessary by the 3-cm–exposed electrode. Thus, we decided to perform TAE with Lipiodol to tumors > 2 cm to delineate the border of the tumors at the CT scan for treatment evaluation after ablation. We tried to make sure that the entire tumor was surrounded as completely as possible by a nonenhanced area on an evaluation CT scan. It is known that occlusion of arterial flow during RFA significantly enlarges the zone of coagulation.38, 39 Our results for local tumor progression rate may be contributed by the effect although the exact impact of this confounding factor is precisely unknown when RFA is performed > 7 days after embolization (as we did) compared with when it is performed immediately (within 3 days) after embolization.

Louha et al.40, 41 reported on the likelihood of spreading cancer cells into peripheral blood by liver resection, ethanol injection, and transarterial embolization. However, the risk of hematogenous dissemination of HCC cells is negligible, because most patients had three or fewer nodules at HCC recurrence, and extrahepatic metastasis was rare in our patient series.

RFA can be repeated in patients with tumor recurrence as long as liver function is relatively intact. The finding that the majority of patients with tumor recurrence underwent RFA indicates that the initial RFA was not so invasive as to seriously damage liver function. As a locoregional therapy, RFA does not prevent de novo carcinogenesis in the remnant liver with chronic viral infection. Another strategy including viral eradication is needed to prevent overall tumor recurrence.42

The prognosis of patients with HCC is dependent mainly on tumor characteristics and liver function also in case of RFA. Patients with severely impaired liver function may not be eligible for RFA. The poorer survival among Child–Pugh Class B and C patients indicates that other strategies including OLT should be considered for longer survival.

RFA is a safe and effective method with satisfactory curability at least locally, and it can be repeated against tumor recurrence. The 5-year survival rate may be better than that previously reported for ethanol injection11, 43 and is obviously better than that of natural course.44, 45 Thus, RFA can be the first choice in the treatment of small HCC, although further follow-up will be necessary for the assessment of longer survival.