Intervention Review

You have free access to this content

Radiofrequency (thermal) ablation versus no intervention or other interventions for hepatocellular carcinoma

  1. Sebastian Weis1,*,
  2. Annegret Franke2,
  3. Joachim Mössner1,
  4. Janus C Jakobsen3,
  5. Konrad Schoppmeyer4

Editorial Group: Cochrane Hepato-Biliary Group

Published Online: 19 DEC 2013

Assessed as up-to-date: 19 SEP 2012

DOI: 10.1002/14651858.CD003046.pub3


How to Cite

Weis S, Franke A, Mössner J, Jakobsen JC, Schoppmeyer K. Radiofrequency (thermal) ablation versus no intervention or other interventions for hepatocellular carcinoma. Cochrane Database of Systematic Reviews 2013, Issue 12. Art. No.: CD003046. DOI: 10.1002/14651858.CD003046.pub3.

Author Information

  1. 1

    University of Leipzig, Division of Gastroenterology and Rheumatology Department of Internal Medicine, Neurology and Dermatology, Leipzig, Germany

  2. 2

    Universität Leipzig, Zentrum für Klinische Studien (ZKS) Leipzig, Leipzig, Germany

  3. 3

    Copenhagen Trial Unit, Centre for Clinical Intervention Research, Department 7812, Rigshospitalet, The Cochrane Hepato-Biliary Group, Copenhagen, Sjælland, Denmark

  4. 4

    Euregio-Klinik GmbH, Internal Medicine, Nordhorn, Germany

*Sebastian Weis, Division of Gastroenterology and Rheumatology Department of Internal Medicine, Neurology and Dermatology, University of Leipzig, Liebigstrasse 20, Leipzig, 04103, Germany. sebastian.weis@medizin.uni-leipzig.de.

Publication History

  1. Publication Status: New search for studies and content updated (conclusions changed)
  2. Published Online: 19 DEC 2013

SEARCH

 
Characteristics of included studies [ordered by study ID]
Brunello 2008

MethodsRandomised trial.


Participants139 patients: RFA = 70; PEI = 69.

Recruitment: January 2001 to September 2004.

Country: Italy.


InterventionsRFA versus PEI.


OutcomesPrimary outcome: complete response at 1 year.

Secondary outcome: early complete response (at 30-50 days after treatment), complications, survival, costs.


Inclusion Criteria
  • Child-Pugh cirrhosis A and B.
  • 1 to 3 nodules, each ≤ 30 mm.
  • HCC confirmed by radiology or histology.


Exclusion Criteria
  • no liver cirrhosis.
  • Child-Pugh cirrhosis C.
  • platelet count < 40,000 mm3.
  • INR > 1.75.
  • prothrombin time > 40 seconds.
  • hypovascular HCC.
  • lesions not detectable by ultrasound.
  • lesions close to gallbladder, hilum, colon, stomach (< 1 cm).
  • venous invasion.
  • metastasis.
  • tumour > 30 mm.
  • patients suitable for liver transplantation or resection.


Follow-up26 months.


NotesPatients suitable for resection were excluded.

Patients with incomplete response (2 cycles) were further treated with PEI, RFA, or TACE.

Intention-to-treat analysis for complete response and survival was performed.

Trial stopped early after conclusive interim analysis indicating effects that exceeded the expectations.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskSealed envelopes.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskNumber and reasons for drop-outs were reported.

Selective reporting (reporting bias)Low riskAll outcomes were appropriately reported.

Other biasUnclear riskAn interim analysis was carried out after 3 years leading to the interruption of the study, because the difference in 1-year complete response rate (the primary endpoint) was larger than expected. It remains unclear, whether this was a preplanned analysis. In addition, the impact on survival (a secondary endpoint) remains speculative.

More patients received a second treatment cycle in the PEI group.

Vested interest bias?Low riskNo previous RFA trial was performed by this group. Industry or other funding bias was ruled out.

Overall risk of bias assessmentHigh riskHigh risk of bias (other bias unclear: number of second treatment cycles; baseline imbalance unclear; interim analysis not mentioned as preplanned).

Chen 2006

MethodsRandomised trial.


Participants180 patients: partial hepatectomy = 90; percutaneous local ablative therapy = 90.

Recruitment: November 1999 to June 2004.

Country: China.


InterventionsRFA versus hepatic resection.


OutcomesPrimary outcome: tumour recurrence rate at 2 years.

Secondary outcomes: overall survival, tumour-free survival, pain, length of hospital stay.


Inclusion Criteria
  • HCC confirmed histologically or by 2 imaging procedures + AFP > 400 ng/mL.
  • 1 HCC nodule ≤ 5 cm.
  • aged 18 to 75 years.
  • no extrahepatic metastasis.
  • no radiological evidence of invasion into major portal/hepatic vein branches.
  • Child-Pugh cirrhosis A with no history of encephalopathy, ascites refractory to diuretics, or variceal bleeding.
  • indocyanine green retention at 15 minutes < 30%.
  • platelet count > 40,000 mm3.
  • suitable for surgical or percutaneous local ablative therapy treatment.


Exclusion Criteria
  • none quoted.


Follow-upRFA 27.9 ± 10.6 months, resection 29.2 ± 11.9 months


Notes19 patients who were randomised to RFA withdrew their consent and received resection. Survival analysis according to the intention-to-treat principle.

When residual tumour was present, RFA or PEI and TACE were performed. Only 62% of patients were treated with RFA once.

2 patients randomised to surgery were treated with PEI because of disseminated disease.

Focal nodular hyperplasia was found in 1 patient instead of HCC.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskComputer at a central registry.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs were reported.

Selective reporting (reporting bias)Low riskAll outcomes were appropriately reported.

Other biasUnclear riskA high percentage of patients randomised to the percutaneous local ablative therapy arm withdrew their consent and received surgical resection. They were analysed in an intention-to-treat analysis.

Vested interest bias?Low riskNo previous RFA trial was performed by this group. Industry or other funding bias was ruled out.

Overall risk of bias assessmentHigh riskHigh risk of bias (other bias: > 20% withdrawals in the percutaneous local ablative therapy arm; baseline imbalance; source of funding).

Feng 2012

MethodsProspective randomised trial.


Participants168 patients: RFA = 84; surgical resection = 84.

Recruitment: January 2005 to March 2008.

Country: China.


InterventionsRFA versus hepatic resection.


OutcomesPrimary outcome: 3-year survival.

Secondary outcomes: 1- and 2- year survival, tumour recurrence, complications.


Inclusion Criteria
  • diagnosis of HCC confirmed at study authors' hospital.
  • no more than 2 nodules.
  • tumour diameter < 4 cm.
  • Child–Pugh cirrhosis A or B.
  • no intrahepatic and extrahepatic metastases.
  • no invasion of the portal vein.
  • no invasion of the hepatic vein trunk or secondary branches.
  • indocyanine green retention at 15 minutes < 30%.
  • platelet count > 50,000 mm3.
  • prolonged prothrombin time of < 5 seconds.
  • no other previous anti-tumour therapy.
  • suitable candidates for both hepatic resection and RFA.


Exclusion Criteria
  • patients met the inclusion criteria but declined to participate.
  • severe portal hypertension.
  • history of oesophageal variceal haemorrhage.
  • severe hypersplenism syndrome.
  • refractory ascites.
  • patients whose permanent pathology after treatment suggested metastatic
  • liver cancer or primary liver cancer of another tissue type.
  • patients who were willing to receive a liver transplantation


Follow-up36 months.


NotesThe absolute number of deaths that occurred in the resection group were 18 including 1 patient died of a cerebral vascular accident. 24 patients died in the RFA group including one patient who died in a motor vehicle accident.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandom number generator.

Allocation concealment (selection bias)Low riskSealed envelopes.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs reported.

Selective reporting (reporting bias)Low riskPower calculation performed. Intention-to-treat analysis performed.

Other biasLow riskZelen method used for stratification. Written informed consent signed after agreement with randomisation.

Patients who died of causes unrelated to HCC were judged as lost to follow-up.

Vested interest bias?Low riskNo previous RFA trial was performed by this group. Industry or other funding bias was ruled out.

Overall risk of bias assessmentLow riskHigh risk of bias due to unclear other bias.

Ferrari 2007

MethodsRandomised trial.


Participants81 patients: RFA = 40; laser ablation = 41.

Recruitment: January 2003 to December 2005.

Country: Italy.


InterventionsRFA versus laser ablation.


OutcomesSurvival, complications, complete tumour ablation, disease recurrence.


Inclusion Criteria
  • histologically confirmed HCC.
  • single nodule ≤ 40 mm or no more than 3 nodules ≤ 30 mm each.
  • no previous HCC treatment.
  • performance status ECOG 0-2.
  • cardiac and pulmonary function WHO 0-2.
  • aged 19 to 82 years.
  • underlying liver cirrhosis.
  • delayed inclusion.
  • prothrombin activity < 40%.
  • INR > 1.99.
  • platelet count > 40,000 mm3.


Exclusion Criteria
  • tumour invasion into vessels or main bile duct.
  • extrahepatic metastases.
  • Child-Pugh C cirrhosis.
  • severely decompensated cirrhosis.
  • major oesophageal varices.


Follow-upNot stated.


NotesNo primary and secondary endpoints were defined.

The study authors stated that there were no complications at all.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskComputer randomisation.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported.

Selective reporting (reporting bias)Low riskAll relevant outcomes were reported.

Other biasUnclear riskOutcomes not defined.

Unclear whether normalisation of coagulation did improve Child-Pugh status.

Vested interest bias?Unclear riskNo previous RFA trial was performed by this group. Industry or funding bias could not be ruled out.

Overall risk of bias assessmentHigh riskHigh risk of bias.

Giorgio 2011

MethodsRandomised trial.


Participants285 patients: RFA = 142; PEI = 143.

Recruitment: January 2005 to January 2010.

Country: Italy.


InterventionsRFA versus PEI.


OutcomesPrimary outcome: 5-year survival, rate of recurrence.

Secondary survival: feasibility of both procedures considering segmental location.


Inclusion Criteria
  • single nodule ≤ 3 cm.
  • Child-Pugh cirrhosis A or B.
  • no ascites.
  • no portal vein or segmental portal thrombosis.


Exclusion Criteria
  • > 1 nodule.
  • > 3 cm in diameter.
  • Child-Pugh C cirrhosis.
  • patients eligible for resection or liver transplantation.
  • patients with extrahepatic diseases.


Follow-up8 to 68 months.


NotesAdditional information from study authors was obtained: HR of overall survival.

14 patients randomised to RFA could not be treated by RFA and were not included in the survival analysis.

Patients and field staff were blinded to treatment assignment.

In addition, the total number of deaths were obtained from the study authors and were: HCC < 3 cm treated with PEI 45; HCC < 3 treated with RFA 30; HCC < 2 cm treated with PEI 14; HCC < 2 cm treated with RFA 11.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandom number generator.

Allocation concealment (selection bias)Low riskCoded list.

Blinding of survival outcomesLow riskPatients and field staff blinded to treatment.

Incomplete outcome data (attrition bias)
All outcomes
Unclear riskDrop-outs reported, but 14 patients randomised to RFA were treated with PEI and not included in the survival analysis.

Selective reporting (reporting bias)Low riskAll relevant outcomes were reported. HR was given only for overall survival.

Other biasLow riskThere was no other source of bias detected.

Vested interest bias?Low riskVested interest bias such as academic bias or industry funding, etc. was not present.

Overall risk of bias assessmentHigh riskHigh risk of bias.

Huang 2010

MethodsRandomised trial.


Participants230 patients: RFA 115; surgical resection 115.

Recruitment: March 2003 to January 2005.

Country: China.


InterventionsRFA versus hepatic resection.


OutcomesPrimary outcome: overall survival.

Secondary outcome: recurrence-free survival, overall recurrence, adverse events, length of hospital stay.


Inclusion Criteria
  • diagnosis of HCC confirmed in the study hospital.
  • tumour within the Milan criteria.
  • no extrahepatic metastasis or obvious vascular invasion.
  • Child-Pugh cirrhosis A or B.
  • no previous or simultaneous malignancies.
  • indocyanine green retention at 15 minutes < 20%.
  • platelet count > 50,000 mm3 or correctable by transfusion.
  • prothrombin time < 5 seconds.
  • HBV-DNA quantification in HBV-infected patients < 10,000 copies/mL.
  • no previous HCC treatment.
  • suitable to be treated by resection or RFA.


Exclusion Criteria
  • severe portal hypertension with history of oesophageal variceal haemorrhage, with large oesophageal varices, refractory ascites.
  • willing to receive liver transplantation.


Follow-up1 to 60 months.


NotesStudy authors reported limitations of the trial. Patients that were willing to receive liver transplantation were excluded.

Subgroup analysis regarding solitary HCC < 3 cm, solitary HCC 3 to 5 cm and multifocal HCC.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskSealed envelopes.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influenced by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs and protocol violations were reported.

Selective reporting (reporting bias)Low riskAll relevant outcomes were reported.

Other biasLow riskNo other source of bias.

Vested interest bias?Low riskNo previous RFA trial was performed by this group. Industry or other funding bias was ruled out.

Overall risk of bias assessmentLow riskLow risk of bias regarding efficacy outcomes.

Lencioni 2003

MethodsRandomised controlled trial.


Participants102 patients: RFA = 52; PEI = 50.

Recruitment: unclear.

Country: Italy, Germany.


InterventionsRFA versus PEI.


OutcomesPrimary outcome: overall survival.
Secondary outcome: local recurrence free and event free (metastasis, new HCC) survival, adverse events.


Inclusion Criteria
  • adults with cirrhosis and (a) single HCC ≤ 5 cm in diameter or (b) 3 HCCs each ≤ 3 cm in diameter.
  • HCCs all located at least 1 cm from the hepatic hilum or the gall bladder.
  • absence of vascular invasion or extrahepatic metastases.
  • Child-Pugh cirrhosis A or B.
  • prothrombin activity > 50%.
  • platelet count > 50,000 mm3
  • no previous treatment for HCC.
  • ineligibility for surgical resection or transplantation.
  • written informed consent.


Exclusion Criteria
  • none given.


Follow-upRFA mean 22.9 months, PEI mean 22.4 months.


Notes2 patients randomised to RFA were excluded due to violation of the inclusion criteria (1 with a large carcinoma > 5 cm and 1 with extrahepatic tumour).

Histological confirmation of HCC was only obtained in 45/102 patients.

Based on contact with the statistician of the trial we have been informed that the risk ratios are indeed HRs.

No sample size calculation reported.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskCentral registry.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskReasons for drop-out/withdrawal stated.

Selective reporting (reporting bias)Low riskAll relevant outcomes were reported.

Other biasLow riskNo other source of bias.

Vested interest bias?Low riskNo previous randomised controlled trials but retrospective studies published. Industry or other funding bias was excluded.

Overall risk of bias assessmentLow riskLow risk of bias regarding efficacy outcomes.

Lin 2004 (RFA vs. PEI)

MethodsRandomised trial.


Participants157 patients: RFA 52; high-dose PEI 53; PEI 52.

Recruitment: April 2000 to April 2002.

Country: Taiwan.


InterventionsRFA versus high-dose PEI versus PEI.

Comparison 1: RFA versus PEI.


OutcomesPrimary outcome: local tumour progression.

Secondary outcome:complete tumour necrosis, overall survival, cancer-free survival.


Inclusion Criteria
  • histologically or cytologically confirmed HCC.
  • 1 to 3 nodules up to 4 cm.
  • Child-Pugh cirrhosis A or B.


Exclusion Criteria
  • Child-Pugh cirrhosis C.
  • previous HCC treatment.
  • tumour within 5 mm of liver hila or common bile duct.


Follow-up4 to 43 months.


Notes6 patients with PEI, 3 patients with high-dose PEI, and 2 patients with RFA did not receive a complete treatment course or did not achieve complete necrosis and underwent TACE.

Patients of both PEI arms treated as outpatients as long as no adverse event occurred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskLocked computer.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs were reported.

Selective reporting (reporting bias)Low riskAll relevant outcomes were reported.

Other biasLow riskNo source of other bias.

Vested interest bias?Low riskThe same research group performed another randomised clinical trial at the same time. Patients were assigned to 1 of the parallel trials depending on hospital admission during alternative weeks.

Previous retrospective trial on PEI for HCC.

Overall risk of bias assessmentLow riskLow risk of bias regarding efficacy outcomes.

Lin 2004 (RFA vs. PEI-hd)

MethodsRandomised trial.


Participants157 patients: RFA = 52; high-dose PEI = 53; PEI = 52.

Recruitment: April 2000 to April 2002.

Country: Taiwan.


InterventionsRFA versus high-dose PEI versus PEI.

Comparison 2: RFA versus high-dose PEI.


OutcomesPrimary outcome: local tumour progression.

Secondary outcome: complete tumour necrosis, overall survival, cancer-free survival.


Inclusion Criteria
  • histologically or cytologically confirmed HCC.
  • 1 to 3 nodules up to 4 cm.
  • Child-Pugh cirrhosis A or B.


Exclusion Criteria
  • Child-Pugh cirrhosis C.
  • previous HCC treatment.
  • tumour within 5 mm of liver hila or common bile duct.


Follow-up4 to 43 months.


Notes6 patients with PEI, 3 patients with high-dose PEI, and 2 patients with RFA did not receive a complete treatment course or did not achieve complete necrosis and received TACE.

Patients of both PEI arms treated as outpatients as long as no adverse event occurred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk

Allocation concealment (selection bias)Low risk

Blinding of survival outcomesLow risk

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs were reported.

Selective reporting (reporting bias)Low risk

Other biasLow risk

Vested interest bias?Low risk

Overall risk of bias assessmentLow risk

Lin 2005 (RFA vs. PAI)

MethodsRandomised trial.


Participants187 patients: RFA = 62; PEI = 62; PAI = 63.

Recruitment: April 2000 to June 2002.

Country: Taiwan.


InterventionsRFA versus PEI versus PAI.

Comparison 1: RFA versus PAI.


OutcomesPrimary outcome: local recurrence.

Secondary outcome: overall survival, cancer-free survival.


Inclusion Criteria
  • 1 to 3 HCC nodules each ≤ 3 cm.
  • > 1 cm distance to hepatic hilum.
  • absence of extrahepatic 3 cm metastasis or vascular invasion.
  • liver cirrhosis Child A or B.
  • prothrombin time 3 seconds less than control values
  • platelet count > 50,000/mm3.
  • no previous treatment.


Exclusion Criteria
  • Child-Pugh cirrhosis C.
  • previous HCC treatment.
  • tumour located within 1 cm of the liver hilum or common bile duct.


Follow-up41 to 44 months.


Notes
  • 4.8% major complications in RFA, 0% in the PAI and PEI groups.
  • 11 patients had a treatment failure and were treated with additional therapies.
  • Planned early stopping if there was a survival benefit for 1 of the treatments as evaluated every 3 months.
  • patients of PEI/PAI arms treated as outpatients as long as no adverse events occurred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk

Allocation concealment (selection bias)Low risk

Blinding of survival outcomesLow risk

Incomplete outcome data (attrition bias)
All outcomes
Low risk

Selective reporting (reporting bias)Low risk

Other biasLow risk

Vested interest bias?Low risk

Overall risk of bias assessmentLow risk

Lin 2005 (RFA vs. PEI)

MethodsProspective randomised trial.


Participants187 patients: RFA = 62; PEI = 62; PAI = 63.

Recruitment: April 2000 to June 2002.

Country: Taiwan.


InterventionsRFA versus PEI versus PAI.

Comparison 1: RFA versus PEI.


OutcomesPrimary outcome: local recurrence.

Secondary outcome: overall survival, cancer-free survival.


Inclusion Criteria
  • 1 to 3 HCC nodules each ≤ 3 cm.
  • > 1 cm distance to hepatic hilum.
  • absence of extrahepatic metastasis or vascular invasion.
  • Child-Pugh cirrhosis A or B.
  • prothrombin time 3 seconds less than control values.
  • platelet count > 50,000/mm3.
  • no previous treatment.


Exclusion Criteria
  • Child-Pugh cirrhosis C.
  • previous HCC treatment.
  • tumour located within 1 cm of the liver hilum or common bile duct.


Follow-up4 to 44 months.


Notes
  • 4.8% major complications in RFA, 0% in the PAI and PEI groups.
  • 7 patients had a treatment failure and were treated with additional therapies.
  • Planned early stopping if there was a survival benefit for 1 of the treatments as evaluated every 3 months.
  • patients of PEI/PAI arms treated as outpatients as long as no adverse events occurred.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskComputer randomisation.

Allocation concealment (selection bias)Low riskLocked computer.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskDrop-outs did not occur.

Selective reporting (reporting bias)Low riskAll outcomes were addressed.

Other biasLow riskLocal recurrence was primary outcome, not survival, thus sample size calculation was based on local recurrence.

Vested interest bias?Low riskThe same trial group performed another randomised clinical trial at the same time. Patients were assigned to 1 of the parallel trials depending on hospital admission during alternative weeks.

Previous retrospective trial on PEI for HCC.

Overall risk of bias assessmentLow riskLow risk of bias regarding efficacy outcomes.

Shibata 2002

MethodsRandomised trial.


Participants72 patients: RFA = 36; percutaneous microwave ablation = 36.

Recruitment: March 1999 to October 2000.

Country: Japan.


InterventionsRFA versus percutaneous microwave coagulation.


OutcomesPrimary outcome: tumour recurrence.
Secondary outcome: pain, complications, duration of the procedure.


Inclusion Criteria
  • HCC diagnosis confirmed by ultrasound-guided needle biopsy.
  • a solitary HCC nodule ≤ 4 cm in diameter or 2 or 3 HCC nodules ≤ 3 cm in diameter.
  • informed consent.


Exclusion Criteria
  • none given.


Follow-up6 to 27 months.


NotesNo data on survival were reported.

Detailed description of complications was included.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskNot reported.

Allocation concealment (selection bias)Low riskSealed envelope.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Unclear riskDrop-outs not reported. No intention-to-treat analysis. Apparently all patients received the assigned treatment.

Selective reporting (reporting bias)High riskSurvival was not a defined outcome and not reported.

Other biasLow riskNo other source of bias.

Vested interest bias?Unclear riskNo previous RFA trial was performed by this group. Industry or funding bias could not be ruled out.

Overall risk of bias assessmentHigh riskHigh risk of bias.

Shiina 2005

MethodsRandomised trial.


Participants232 patients: RFA = 118; PEI = 114.

Recruitment: April 1999 to January 2001.

Country: Japan.


InterventionsRFA versus PEI.


OutcomesPrimary outcome: overall survival.

Secondary outcome: overall recurrence, local tumour progression.


Inclusion Criteria
  • histopathologically or radiologically confirmed HCC.
  • lesions were unresectable or the patient had refused surgery.
  • 1 to 3 HCC nodules each ≤ 3 cm.
  • Child-Pugh cirrhosis A or B.
  • no extrahepatic metastasis or vascular invasion; no previous or simultaneous malignancies.


Exclusion Criteria
  • refractory ascites.
  • platelet count < 50,000/mm3
  • prothrombin activity below < 50%.


Follow-up0.6 to 4.3 years (median 3.1 years).


NotesNo patients were excluded for lesion location.

Detection of fourth lesion in 3 RFA patients and 4 PEI patients.

Intention-to-treat analysis performed.

More experienced physician performed the difficult lesions.

Adverse events reported.

All patients were treated as inpatients until computed tomography stated treatment success.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandom number generator.

Allocation concealment (selection bias)Low riskComputer.

Blinding of survival outcomesLow riskSurvival outcomes may not be significantly influences by a lack of blinding.

Incomplete outcome data (attrition bias)
All outcomes
Low riskNo patient was lost to follow-up. An intention-to-treat analysis was performed.

Selective reporting (reporting bias)Low riskAll outcomes reported.

Other biasLow riskThere was no other bias.

Vested interest bias?Low riskNo previous RFA trial was performed by this group. Industry or other funding bias was ruled out.

Overall risk of bias assessmentLow riskLow risk of bias regarding efficacy outcomes.

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Akamatsu 2004Randomised trial (using sealed envelopes) that was not included due to the following reasons: patients were switched from PEI to RFA as soon as this technique was available. Randomisation of 6 versus 10 patients. Therefore, the study design was considered insufficient to evaluate RFA effects.

Amarnath 2006Retrospective review.

Buscarini 1996Not a controlled trial.

Chen 2005Randomisation not clear. Study authors did not reply to our contact approaches via email.

Cheng 2008Publication retracted in May 2009.

Cho 2010Not a controlled trial. Markov model analysis.

Cuschieri 1999Not a controlled trial.

Gan 2004Randomisation not clear. Study authors did not reply.

Giorgio 2011aRFA also used to treat portal vein thrombosis.

Goldberg 1998Not a controlled trial.

Gory 2012Retrospective multicentre analysis.

Jiao 1999Not a controlled trial.

Khan 2007Not a randomised trial.

Kobayashi 2007RFA compared with combination treatment.

Kurokohchi 2005Not a randomised trial.

Livraghi 1999Controlled trial without randomisation.
Different diagnostic follow-up schedules in the 2 treatment groups.
No data concerning overall survival.

Lü 2006Radiofrequency or microwave ablation therapy applied in 1 group compared with surgical resection in the other group. Therefore, RFA effects were not extractable.

Morimoto 2010RFA compared with combination treatment.

Morimoto 2011RFA compared with combination treatment.

Ohmoto 2009Not a controlled trial.

Opocher 2010Retrospective cohort study.

Peng 2012Only recurrent HCC treated.

Santambrogio 2012Not a percutaneous intervention.

Sreenivasan 2010Retrospective cohort study.

Ueno 2009Retrospective cohort study.

Zhang 2005Quasi-randomised trial.

Zhang 2007RFA compared with combination treatment.

 
Characteristics of studies awaiting assessment [ordered by study ID]
Di Costanzo 2011

MethodsRandomised clinical trial.

Participants80 patients: RFA = 40; laser ablation = 40.

Recruitment: January 2009 to October 2010.

Country: Italy.

InterventionsRFA versus laser ablation.

OutcomesPrimary outcome: complete tumour ablation, time to recurrence.

Secondary outcome: overall survival.

NotesFrom the meeting abstract. Insufficient information to judge bias, randomisation process, etc. Author contact failed.

Kuansheng 2011

MethodsRandomised clinical trial.

Participants168 patients: RFA = 84; surgical resection = 84.

Recruitment: December 2004 to October 2007

Country: most likely China.

InterventionsRFA versus surgical resection.

Outcomes1-, 2- and 3-year survival.

NotesFrom the meeting abstract. Insufficient information to judge bias, randomisation process, etc. Probably data published in the trial by Feng et al (Feng 2012).

 
Characteristics of ongoing studies [ordered by study ID]
NCT00814242

Trial name or titleHepatectomy Versus Radiofrequency Ablation for Hepatocellular Carcinoma Adjacent to Major Blood Vessels.

MethodsRandomisation unclear.

Participants
  • approximately 120 participants.


Inclusion:

  • HCC at complex site (see below).
  • Child-Pugh cirrhosis.
  • no dysfunction in major organs; blood routine, kidney function, cardiac function, and lung function are basically normal.
  • aged 18 to 70 years.

InterventionsRFA versus surgery.

OutcomesPrimary:

  • tumour recurrence rate after 1 and 2 years.
  • disease-free survival.
  • overall survival.


Secondary:

  • overall survival rate in 1, 2, 3, or 5 years.
  • disease-free survival in 1, 2, 3, or 5 years.
  • hepatic function of patients after surgery.
  • complications and the decline level of serum AFP concentration.

Starting dateDecember 2008

Contact informationFeng Shen, MD, Study Chair.

Lehua Shi, MD, telephone: 0086-21-25070784.
Email: shilh@ehbh.cn.

Yong Xia, MD, telephone: 0086-21-25074943.
Email: xiay99@hotmail.com.

NotesHepatoma at a complex site (8 segment of liver, caudate lobe, the substantial depth of liver below hepatic integument adjacent to the trunk of inferior vena cava, hepatic vein, and portal vein).

 
Comparison 1. Overall survival

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Overall survival: control interventions versus RFA12Hazard Ratio (Random, 95% CI)Subtotals only

    1.1 Resection versus RFA
3578Hazard Ratio (Random, 95% CI)0.71 [0.44, 1.15]

    1.2 PEI or PAI versus RFA
81088Hazard Ratio (Random, 95% CI)1.64 [1.31, 2.07]

    1.3 Laser ablation versus RFA
181Hazard Ratio (Random, 95% CI)1.62 [0.62, 4.22]

 2 Fixed-effect model3Hazard Ratio (Fixed, 95% CI)0.76 [0.58, 1.00]

 3 Trials with low risk of bias2Hazard Ratio (Random, 95% CI)0.56 [0.40, 0.78]

 4 Resection versus RFA, dichotomous outcome2398Risk Ratio (M-H, Random, 95% CI)0.60 [0.44, 0.82]

 5 PEI or PAI versus RFA81088Hazard Ratio (Random, 95% CI)1.64 [1.31, 2.07]

    5.1 PEI versus RFA
7994Hazard Ratio (Random, 95% CI)1.62 [1.27, 2.07]

    5.2 PAI versus RFA
194Hazard Ratio (Random, 95% CI)1.82 [0.94, 3.55]

 
Comparison 2. Two-year survival

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Control interventions versus RFA9Hazard Ratio (Random, 95% CI)Subtotals only

    1.1 Resection versus RFA
3578Hazard Ratio (Random, 95% CI)0.51 [0.24, 1.08]

    1.2 PEI/PAI versus RFA
6715Hazard Ratio (Random, 95% CI)1.82 [1.34, 2.47]

 2 Resection versus RFA, low risk of bias trials2398Hazard Ratio (Random, 95% CI)0.38 [0.17, 0.84]

 3 PEI or PAI versus RFA, low risk of bias trials5576Hazard Ratio (Random, 95% CI)1.98 [1.41, 2.78]

 
Comparison 3. Event-free survival

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Event-free survival: control interventions versus RFA11Hazard Ratio (Random, 95% CI)Subtotals only

    1.1 PEI/PAI versus RFA
7949Hazard Ratio (Random, 95% CI)1.55 [1.31, 1.85]

    1.2 Resection versus RFA
3578Hazard Ratio (Random, 95% CI)0.70 [0.54, 0.91]

    1.3 Laser ablation versus RFA
181Hazard Ratio (Random, 95% CI)1.20 [0.50, 2.89]

 2 PEI or PAI versus RFA, low risk of bias trials6678Hazard Ratio (Fixed, 95% CI)1.59 [1.33, 1.90]

 
Comparison 4. Local progression/recurrences

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Local progression: control intervention versus RFA9Hazard Ratio (Random, 95% CI)Subtotals only

    1.1 PEI/PAI versus RFA
6678Hazard Ratio (Random, 95% CI)2.44 [1.71, 3.49]

    1.2 Resection versus RFA
1168Hazard Ratio (Random, 95% CI)0.48 [0.28, 0.82]

    1.3 Microwave ablation versus RFA
172Hazard Ratio (Random, 95% CI)2.14 [0.67, 6.80]

    1.4 Laser ablation versus RFA
181Hazard Ratio (Random, 95% CI)1.12 [0.40, 3.09]

 
Comparison 5. Other secondary outcomes

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Rate of complications: RFA versus other interventions12Odds Ratio (M-H, Random, 95% CI)Subtotals only

    1.1 Resection versus RFA
3559Odds Ratio (M-H, Random, 95% CI)8.24 [2.12, 31.95]

    1.2 PEI/PAI versus RFA
81088Odds Ratio (M-H, Random, 95% CI)0.70 [0.33, 1.48]

    1.3 Microwave ablation versus RFA
172Odds Ratio (M-H, Random, 95% CI)4.38 [0.46, 41.22]

 2 Rate of complications: PEI or PAI versus RFA81088Odds Ratio (M-H, Random, 95% CI)0.70 [0.33, 1.48]

    2.1 PEI versus RFA
7994Odds Ratio (M-H, Random, 95% CI)0.79 [0.37, 1.72]

    2.2 PAI versus RFA
194Odds Ratio (M-H, Random, 95% CI)0.09 [0.00, 2.00]

 3 PEI/PAI versus RFA, low risk of bias trials6Odds Ratio (M-H, Random, 95% CI)Subtotals only

 4 Length of hospital stay; control interventions versus RFA (days)8Std. Mean Difference (IV, Random, 95% CI)Subtotals only

    4.1 Resection versus RFA
3559Std. Mean Difference (IV, Random, 95% CI)2.18 [1.97, 2.39]

    4.2 PEI/PAI versus RFA
5576Std. Mean Difference (IV, Random, 95% CI)-1.50 [-3.69, 0.68]

 5 Length of hospital stay; PEI or PAI versus RFA (days)5576Std. Mean Difference (IV, Random, 95% CI)-1.50 [-3.69, 0.68]

    5.1 PEI versus RFA
4482Std. Mean Difference (IV, Random, 95% CI)-1.47 [-4.23, 1.30]

    5.2 PAI versus RFA
194Std. Mean Difference (IV, Random, 95% CI)-1.66 [-2.16, -1.17]

 
Comparison 6. PEI or PAI versus RFA, overall survival, other

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 PEI versus RFA, dichotomous outcome6986Risk Ratio (M-H, Random, 95% CI)1.76 [0.97, 3.22]

 2 PEI/PAI versus RFA, low risk of bias trials6Hazard Ratio (Random, 95% CI)1.93 [1.46, 2.55]

 3 PEI versus RFA, Italian versus Asian trials6915Hazard Ratio (Random, 95% CI)1.60 [1.24, 2.08]

    3.1 Italian trials
3512Hazard Ratio (Random, 95% CI)1.24 [0.84, 1.84]

    3.2 Asian trials
3403Hazard Ratio (Random, 95% CI)1.95 [1.38, 2.75]

 
Summary of findings for the main comparison. Hepatic resection compared with radiofrequency ablation

Hepatic resection compared with radiofrequency ablation for early hepatocellular carcinoma

Patient or population: 578 patients with early HCC

Intervention: hepatic resection

Comparison: RFA

OutcomesHazard ratio (95% CI)No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Overall survival0.71 (0.44 to 1.15)578

(3)
+++O
moderate
-

2-year survival-0.51 (0.24 to 1.08)578

(3)
+++O
moderate
-

Event-free survival0.70 (0.54 to 0.91)578

(3)
+++O
moderate
-

Local progression0.48 (0.28 to 0.82)168

(1)
⊕⊕⊝⊝
low
Only 1 trial reported on local progression.

Rate of complications8.24 (2.12 to 31.95)578

(3)
++++
high
Statistical method is: odds ratio (M-H, random, 95% CI)

Length of hospital stay (days)2.18 (1.97 to 2.39)578

(3)
++++
high
Statistical method is: standard. mean difference (IV, random, 95% CI)

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

 CI: confidence interval; HCC: hepatocellular carcinoma; HR: hazard ratio; PAI: percutaneous acetic acid injection; PEI: percutaneous ethanol injection; RFA: radiofrequency ablation.
 
Summary of findings 2. Percutaneous ethanol injection compared with radiofrequency ablation for early hepatocellular carcinoma

Percutaneous ethanol injection compared with radiofrequency ablation for early hepatocellular carcinoma

Patient or population: 1088 patients with early HCC

Intervention: PEI/PAI

Comparison: RFA

OutcomesHazard ratio (95% CI)No of participants
(studies)
Quality of the evidence
(GRADE)
Comments

Overall survival1.64 (1.31 to 2.07)1088

(6)
⊕⊕⊕⊝
moderate
As there were 2 three-armed trials, we report 6 trials with 8 comparisons.

2-year survival1.82 (1.34 to 2.47)715

(4)
⊕⊕⊕⊝
moderate
-

Event-free survival1.55 (1.31 to 1.85)949

(5)
⊕⊕⊕⊝
moderate
-

Local progression2.44 (1.71 to 3.49)678

(4)
⊕⊕⊕⊝
moderate
-

Rate of complications0.70 (0.33 to 1.48)1088

(6)
⊕⊕⊕⊝
moderate
Statistical method is: odds ratio (M-H, random, 95% CI).

As there were 2 three-armed trials, it is 5 studies with 7 comparisons.

Length of hospital stay-1.50 (-3.69 to 0.68)576

(3)
⊕⊕⊕⊝
moderate
Statistical method is: standard. Mean difference (IV, random, 95% CI).

As there were 2 three-armed trials, it is 3 studies with 5 comparisons.

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

 CI: confidence interval; HCC: hepatocellular carcinoma; HR: hazard ratio; PAI: percutaneous acetic acid injection; PEI: percutaneous ethanol injection; RFA: radiofrequency ablation.
 
Summary of findings 3. Clinical characteristics of patients treated with radiofrequency ablation or percutaneous ethanol injection

Child-Pugh AetiologyTumourOverall survival





Italian Trials ABHBVHCVHBC+HCVAlcoholOthersMean number of nodules1 tumourMean sizeSize < 2 cmHR (95% CI)

Brunello 20080.780.220.040.650.040.140.121.270.782.30.391.14 (0.65 to 2.01)

Giorgio 20110.510.490.410.590.000.000.001.0012.3-1.23 (0.71 to 2.14)

Lencioni 20030.780.220.150.430.250.150.051.390.72.20.255.00 (0.55 to 45.82)

Mean0.690.310.200.560.100.090.051.220.832.270.321.24 (0.84 to 1.83)

East-Asian Trials    

Lin 2004 (RFA vs. PEI)0.770.220.690.29--0.021.310.752.90.292.08 (1.02 to 4.20)

Lin 2005 (RFA vs. PEI)0.750.250.670.31--0.021.260.82.40.62.01 (1.03 to 3.92)

Shiina 20050.730.270.130.81--0.061.630.57-0.441.86 (0.33 to 0.88)

Mean0.750.250.50.47--0.031.400.692.650.521.95 (1.38 to 2.75)

Student's t-testn.s.n.s.n.s.n.s.--n.s.n.s.n.s.n.s.n.s. -

 CI: confidence interval; HBV: hepatitis B virus; HCV: hepatitis C virus; HR: hazard ratio; n.s.: not significant (P value > 0.05); PEI: percutaneous ethanol injection; RFA: radiofrequency ablation.