A SYSTEMATIC REVIEW OF RADIOFREQUENCY ABLATION FOR THE TREATMENT OF LIVER TUMOURS

Transcription

1 ASERNIP S Australian Safety and Efficacy Register of New Interventional Procedures-Surgical A SYSTEMATIC REVIEW OF RADIOFREQUENCY ABLATION FOR THE TREATMENT OF LIVER TUMOURS ASERNIP-S REPORT NO. 28 October 2002 Australian Safety and Efficacy Register of New Interventional Procedures Surgical The Royal Australasian College of Surgeons

2 A SYSTEMATIC REVIEW OF RADIOFREQUENCY ABLATION FOR THE TREATMENT OF LIVER TUMOURS LM Sutherland JAR Williams RTA Padbury DG Gotley B Stokes GJ Maddern ISBN Published October 2002 This report should be cited in the following manner: Sutherland LM, et al. A Systematic Review of Radiofrequency Ablation for the Treatment of Liver Tumours. ASERNIP-S Report No. 28. Adelaide, South Australia: ASERNIP-S, October 2002 Copies of these reports can be obtained from: The Australian Safety and Efficacy Register of New Interventional Procedures - Surgical The Royal Australasian College of Surgeons PO Box 688, North Adelaide, South Australia 5006 AUSTRALIA Fax: College.asernip@surgeons.org

3 The Safety and Efficacy Classification for the Systematic Review of Radiofrequency Ablation for the Treatment of Liver Tumours was ratified by: The ASERNIP-S Management Committee on October 2002 The Council of the Royal Australasian College of Surgeons in October 2002

4 TABLE OF CONTENTS Executive Summary Safety and Efficacy Classification System Review Group Membership 1.0 OBJECTIVE INTRODUCTION Liver Physiology Liver Tumours Treatment Conventional Treatments Radiofrequency Ablation () Summary METHODS Literature Search Protocol Literature Search Strategies Assessment Methods DESCRIPTION OF STUDIES RESULTS DISCUSSION REFERENCE LIST...23 APPENDIX A:...27 Table A.1: Designations of levels of evidence APPENDIX B:...29 Table B.1: Excluded comparative studies...30 Table B.2: Ongoing or unpublished clinical trials...32 APPENDIX C:...33 Table C.1: Study profile and quality assessment...34 Table C.2: Critical appraisal...49 Table C3: Efficacy outcomes for for the treatment of liver tumours RCTs and quasi-rcts.55 Table C4: Safety outcomes for for the treatment of liver tumours RCTs and quasi-rcts...58 Table C5: Efficacy outcomes for for the treatment of liver tumours- Non-randomised comparative studies...59 Table C6: Safety outcomes for for the treatment of liver tumours- Non-randomised comparative studies...61 Table C7: Evaluation of follow-up computed tomography (CT) imaging - Non-randomised comparative studies...62 GLOSSARY:...63 i iii v

5 EXECUTIVE SUMMARY Background The aim of this review was to compare the safety and efficacy of radiofrequency ablation for primary hepatocellular carcinoma or metastatic colorectal liver carcinoma, in comparison to other surgical and non-surgical therapeutic techniques, on the basis of a systematic assessment of the literature. The surgical comparative techniques included resection or hepatic artery infusion chemotherapy. The non-surgical comparative interventions included local ablative therapies such as percutaneous ethanol injection (PEI); cryotherapy; or procedures that produce local heat such as microwave coagulation therapy (MCT) or laser-induced thermotherapy (LITT). Methods Search strategy Studies were identified by searching MEDLINE, PREMEDLINE, PREMEDLINE and MEDLINE, EMBASE, Current Contents, Cochrane Library, Science Citation Index, from inception to week Clinical Trials Database (US), NHS Centre for Research and Dissemination (UK), NHS Health Technology Assessment (UK), National Research Register (UK), EORTC Protocols Database, National of Institute Health (US) and CancerLit (US) were searched on 18/4/2002. This was supplemented by hand-searching recent conference proceedings from specialist societies and conducting internet searches. Additional articles were identified through the reference sections of the studies retrieved. Study selection Randomised controlled trials, quasi-randomised controlled trials and nonrandomised comparative studies assessing patients treated with and either one or more other comparative invention/s were included for review. Patient safety outcomes for radiofrequency ablation were assessed in terms of common end-points reported for surgical treatment or non-surgical ablative treatments, which included major and minor complications. In terms of efficacy, the question was whether radiofrequency ablation produced at least equivalent clinical outcomes, in comparison to surgical treatment or non-surgical ablative treatments. Data collection and analysis Data from the included studies were extracted by the ASERNIP- S Researcher using standardised data extraction tables developed a priori and checked by a second researcher. It was not considered appropriate to pool results across studies. Relative risks (RR) or weighted mean differences (WMD) and the 95% confidence intervals (CI) were calculated individually for the same outcomes in the RCTs and the quasi-rct. Results The evidence was limited by small sample size, short follow-up times and a lack of comparability between the outcome measures. Despite the limitations of the data, generally resulted in larger and more complete areas of ablation and may also be associated with higher survival rates compared to the other ablative techniques assessed in this review. Surgical resection was associated with a lower rate of recurrence and an increased time interval to recurrence compared to. However, these two procedures are usually performed on different patient groups, with being performed on patients who are unable to undergo surgical resection. i

6 Classifications and recommendations Classifications Evidence rating - Average Safety - At least as safe compared to comparator procedure(s) Efficacy - Efficacy cannot be determined Recommendations It was recommended that surgeons practicing radiofrequency ablation for primary hepatocellular carcinoma or metastatic colorectal liver carcinoma should participate in an audit of their outcomes of, preferably at a national level. Important note The information contained in this report is a distillation of the best available evidence located at the time the searches were completed as stated in the protocol. Please consult with your medical practitioner if you have further questions relating to the information provided, as the clinical context may vary from patient to patient. ii

7 THE ASERNIP-S CLASSIFICATION SYSTEM Evidence Rating The evidence for ASERNIP-S systematic reviews is classified as Good, Average or Poor, based on the quality and availability of this evidence. High quality evidence is defined here as having a low risk of bias and no other significant flaws. While high quality randomised controlled trials are regarded as the best kind of evidence for comparing interventions, it may not be practical or ethical to undertake them for some surgical procedures, or the relevant randomised controlled trials may not yet have been carried out. This means that it may not be possible for the evidence on some procedures to be classified as good. Good Most of the evidence is from a high quality systematic review of all relevant randomised trials or from at least one high quality randomised controlled trial of sufficient power. The component studies should show consistent results, the differences between the interventions being compared should be large enough to be important, and the results should be precise with minimal uncertainty. Average Most of the evidence is from high quality quasi-randomised controlled trials, or from nonrandomised comparative studies without significant flaws, such as large losses to follow-up and obvious baseline differences between the comparison groups. There is a greater risk of bias, confounding and chance relationships compared to high-quality randomised controlled trials, but there is still a moderate probability that the relationships are causal. An inconclusive systematic review based on small randomised controlled trials that lack the power to detect a difference between interventions and randomised controlled trials of moderate or uncertain quality may attract a rating of average. Poor Most of the evidence is from case series, or studies of the above designs with significant flaws or a high risk of bias. A poor rating may also be given if there is insufficient evidence. iii

8 Safety and Efficacy Classification Safety At least as safe compared to comparator * procedure(s) This grading is based on the systematic review showing that the new intervention is at least as safe as the comparator. Safety cannot be determined This grading is given if the evidence is insufficient to determine the safety of the new intervention. Less safe compared to comparator * procedure(s) This grading is based on the systematic review showing that the new intervention is not as safe as the comparator. Efficacy At least as efficacious compared to comparator * procedure(s) This grading is based on the systematic review showing that the new intervention is at least as efficacious as the comparator. Efficacy cannot be determined This grading is given if the evidence is insufficient to determine the efficacy of the new intervention. Less efficacious compared to comparator * procedure(s) This grading is based on the systematic review showing that the new intervention is not as efficacious as the comparator. Research Recommendations It may be recommended that an audit or a controlled (ideally randomised) clinical trial be undertaken in order to strengthen the evidence base. Clinical Recommendations Additional recommendations for use of the new intervention in clinical practice may be provided to ensure appropriate use of the procedure by sufficiently qualified/experienced centres and on specific patient types (where appropriate). * A comparator may be the current gold standard procedure, an alternative procedure, a nonsurgical procedure or no treatment (natural history). iv

9 REVIEW GROUP MEMBERS ASERNIP-S Director Professor Guy Maddern Surgical Director ASERNIP-S PO Box 668 NORTH ADELAIDE SA 5006 Protocol Surgeon Mr Tony Williams Gastrointestinal Services Royal Adelaide Hospital North Terrace ADELAIDE SA 5000 Advisory Surgeon Dr Robert Padbury Flinders Medical Centre Flinders Drive BEDFORD PARK SA 5042 Nominated Surgeon Professor David Gotley Mater Medical Centre 293 Vulture Street SOUTH BRISBANE QLD 4101 Other Speciality Surgeon Professor Bryant Stokes St. John of God Hospital 175 Cambridge Street SUBIACO WA 6008 ASERNIP-S Researcher Ms Leanne Sutherland ASERNIP-S PO Box 668 NORTH ADELAIDE SA 5006 v

10 A SYSTEMATIC REVIEW OF RADIOFREQUENCY ABLATION FOR THE TREATMENT OF LIVER TUMOURS 1.0 OBJECTIVE To make recommendations on the safety and efficacy of radiofrequency ablation () for primary hepatocellular carcinoma or metastatic colorectal liver carcinoma, in comparison to other surgical and non-surgical therapeutic techniques, on the basis of a systematic assessment of the peer-reviewed literature. The surgical comparative techniques included resection or hepatic artery infusion chemotherapy. The non-surgical comparative interventions included local ablative therapies such as percutaneous ethanol injection (PEI); cryotherapy; or procedures that produce local heat such as microwave coagulation therapy (MCT) or laserinduced thermotherapy (LITT). 2.0 INTRODUCTION 2.1 Liver Physiology The liver is the largest organ in the body, lying beneath the diaphragm in the right of the abdomen and upper part of the epigastric region. 1 The liver performs multiple complex functions; it secretes bile; is of importance in fat, carbohydrate, and protein metabolism and stores glycogen; it is responsible for the distribution of absorbed nutrients from the blood or gastrointestinal tract and for the elimination of a variety of endogenous and exogenous substrates and toxins. 2 Although the liver constitutes only 4-5% of body weight, it is responsible for 20-25% of the body oxygen consumption and 20% of total energy expenditure. 1 The liver receives a dual blood supply with 75% of flow from the portal vein and 25% from the hepatic artery Liver Tumours Primary liver tumours and liver metastasis from colorectal carcinoma are the two most common malignant tumours to affect the liver. 3,4 Primary liver tumours arise from malignant cells within the liver and hepatocellular carcinoma represents the most common form of primary liver cancer. 3,4 Metastatic liver disease is more common than primary liver disease and develops when malignant cells migrate from other organs to the liver. 5,6 The liver is second only to the lymph nodes as the most common site for metastatic disease. 7 More than half of the patients with metastatic liver disease will die from metastatic complications. 8, Primary Hepatocellular Carcinoma (HCC) Hepatocellular carcinoma (HCC) is one of the most common primary solid cancers in the world 10,11 with an annual incidence estimated to be at least one million new patients 12. HCC is usually associated with liver cirrhosis, either arising from chronic hepatitis B or C, toxic factors, immunologic factors or genetic factors. 13 In most cases HCC is diagnosed in an 1

11 advanced stage of the disease where the prognosis depends on tumour stage, anatomical distribution of tumour lesions and degree of liver function, which will determine the tolerance to invasive treatments. 8,10 Untreated HCC has a median survival of four to six months from the time of diagnosis Metastatic Liver Disease Metastatic liver disease is frequently associated with primary colorectal carcinoma. 15 Approximately 50% of colorectal cancer patients will develop recurrence within five years of initial diagnosis with the liver being the most common site for metastatic disease. 11 For many patients, the progressive involvement of the liver is the primary determinant of long-term survival and without treatment. Patients with colorectal liver metastases have a median survival of six months. 9, Treatment Surgical resection of primary and metastatic liver tumours remains the gold standard of therapy. However, very few patients are suitable for curative surgical resection due to the presence of liver malignancy in unresectable locations, the number and anatomical distribution of tumour lesions, and/or the presence of extrahepatic disease or poor liver function. 17,18 It is estimated that resection is only possible in 20% of patients at time of their diagnosis. 11,19 Furthermore, of those patients who do undergo resection, the 5-year survival rate of patients with HCC or liver metastases is only 20-40%. 20 Thus, for the majority of patients with malignant liver tumours, whether primary or metastatic, who are not candidates for surgical resection, alternative treatment approaches to control and potentially cure the liver disease have been investigated. Chemotherapy infusion through the hepatic artery involves the surgical implantation of a port and the infusion of one or more chemotherapeutic drugs into the liver. Alternatively, ablative techniques treat malignant nodules by direct destruction with either thermal or chemical methods. 2.4 Conventional Treatments Surgical treatments Surgical resection A number of technical and surgical advances have lead to a dramatic impact on resectability rates. It is generally accepted that resection is the only treatment that currently offers a cure, with 5 year survival rates after resection of up to 40%. 21 The number of lesions is no longer considered to be as an important predictor of long-term survival as once thought. The complete excision of all demonstrable tumour with clear resection margins of 1 cm or more has shown to be of much greater importance. 11 If complete excision can be achieved, then the survival rate is similar for single and multiple tumours (up to 8). If complete excision can not be achieved, the surgical resection has no impact on the natural history of the disease. 11 Segmental resection, based on the structure of the liver, allows excision of bilateral or multiple liver lesions that might have previously been deemed unresectable. 2

12 Staged resection is another method by which a large amount of liver parenchyma may be resected without inducing hepatic insufficiency and may be used for bilateral lesions. Hepatic resection of HCC differs from resection for other liver tumours mainly because of the high frequency of cirrhosis. 1 In these situations conservative segmental resection designed to preserve function of the remaining hepatic reserve is used Hepatic artery infusion chemotherapy Hepatic Arterial Infusion Chemotherapy (HAIC), also referred to as regional chemotherapy or transarterial chemoembolisation, is the delivery of chemotherapeutic agents to the liver through a catheter surgically inserted in the hepatic artery, the main pathway through which liver tumours receive their blood supply. 2 By delivering chemotherapy directly to the tumour, HAIC therapy allows for a greater concentration of drug to reach the tumour. This potentially decreases the systemic side effects caused by traditional chemotherapy, due to lower drug exposure to extrahepatic tissues. 11 However careful catheter insertion and after care is crucial if morbidity is to be minimised and real therapeutic benefit achieved Non-surgical treatments Local ablative therapies Percutaneous ethanol injection (PEI) PEI involves a needle being introduced percutaneously into a liver tumour and absolute ethanol or 95% ethanol slowly injected into the lesion. 18 The ethanol is first injected at the deepest aspect of the lesion and the needle withdrawn at small increments to allow uniform diffusion of ethanol within the tumour. 22 Both the needle insertion and injection of ethanol are monitored either by ultrasound, or occasionally by computed tomography. 18,23 As the ethanol diffuses into the cells it induces non-selective protein degradation and cellular dehydration resulting in local areas of coagulation necrosis within and around the tumour. 18 Fibrosis and vascular thrombosis may also contribute to the destruction of the tumour cells. 22 PEI is usually performed in several sessions. 18 It has been suggested that the number of sessions required to ablate the tumour is approximately twice that of the tumour diameter in centimetres Thermal ablative therapies Cryoablation Cryoablation (also known as cryosurgical ablation, CSA) utilises a specifically designed probe to deliver liquid nitrogen to the tissue under the guidance of ultrasound. 25 Tumour cell death is not a direct consequence of lowering tissue temperature, but rather is caused by ice crystal formation during the rapid freezing process. 26 Individual hepatic tumours are frozen sequentially on a tumour-by-tumour basis rather than simultaneously. The tumour is allowed to thaw over a period equivalent to the freezing time before a second freezing process is begun. These multiple freeze-thaw cycles cause destruction to normal cellular structures. 26 Median survival times of months have been reported for patients with colorectal liver metastasis treated with cryotherapy. 27 3

13 Laser-induced thermotherapy (LITT) LITT is also referred to as interstitial laser therapy and is a way of inducing tissue damage and necrosis via heat photocoagulation. 28 The procedure, under either ultrasound or magnetic resonance control 29 utilises a neodymium:yttrium-aluminium-garnet (ND:YAG) laser with exposure times varying from 2 to 30 minutes depending on the tumour size. 30 Microwave coagulation therapy (MCT) MCT (also known as percutaneous microwave coagulation, PMC) relies on the conversion of energy to heat to destroy tissue via a high frequency alternating electromagnetic wave. 31 The electrode is placed into the lesion under ultrasound guidance and microwave treatment applied. 18 During the treatment, water molecules polarise with the alternating electromagnetic wave and heat is generated. Coagulation necrosis and haemostasis results and the reduction in lesion area is reported to be predictable from the length of treatment Radiofrequency Ablation () (also known as radiofrequency thermal ablation, RFTA) is a recently developed thermoablative technique that induces temperature changes by utilising high-frequency alternating current applied via an electrode(s) placed within the tissue to generate ionic agitation. 32 The ionic agitation is generated in the areas surrounding the electrode tip as the ions attempt to change direction and follow the alternating current, thereby creating localised frictional heat. 33 The resultant frictional heating of tissue surrounding the electrode generates localised areas of coagulative necrosis and tissue desiccation. 34 The radiofrequency energy radiates from the individual electrodes into the adjacent tissue. 35 The energy level and thus the heating effect dissipates rapidly at an increasing distance from the electrodes so that the highest temperature will always be at the points nearest to the electrodes. 34 can be applied percutaneously, laparoscopically or intra-operatively Equipment There are three primary types of electrode used for. A brief description of the product design and mechanism of action are provided below RITA Medical Systems This system relies on retractable needle electrodes (Model 70 and 90 Starburst XL Needles, RITA Medical Systems, Mountain View, CA, USA). 36 The needle electrodes consist of an insulated outer needle that houses either 7 (Model 70) or 9 (Model 90) retractable curved electrodes. Four of the electrodes are hollow and are used to measure the temperature of the adjacent tissue. To perform ablation, grounding pads are placed on the patient s thigh, and the tip of the needle with the electrodes retracted is placed in the liver tissue to be treated. The electrodes are partially extended and the alternating current applied with gradually increasing wattage. The temperature at the needle tip is monitored via the electrodes until the temperature reaches between 95 C and 105 C and the electrodes are fully extended. As the tissue begins to 4

14 desiccate, tissue resistance increases and conversely the wattage decreases. Once the ablation cycle is finished, a temperature reading in excess of 50 C from the extended electrodes for one minute is indicative of satisfactory ablation Radiotherapeutics This type of ablation device also has an insulated needle but houses up to 12 solid curved retractable electrodes (LeVeen Needle Electrodes, Radiotherapeutics, Sunnyvale, CA, USA). 37 The tip of the radiofrequency needle is applied to the target tissue. The array of electrodes are fully extended into the tissue such that the electrodes move through the tissue in a constant radius curve from the tip of the needle (generating an umbrella-shaped configuration). An alternating electric current is increased every minute until peak power is attained. The device is designed to respond to changes in tissue impedance during the ablation process rather than tissue temperature. A rise in tissue impedance indicates successful tissue coagulation and ablation Radionics This ablation system employs an insulated hollow needle electrode with an exposed uninsulated needle tip (Cool-Tip Radiofrequency Electrode, Radionics, Burlington, MA, USA). 38 The tip of the single needle or cluster needle electrode (that consists of three of the needles in a parallel triangular cluster) is closed and is able to record the temperature of the adjacent tissue. The shaft of the needle contains two internal channels to allow the needle to be perfused with cool sterile water. The circulating internal water cools the tissue adjacent to the tip of the electrode thus preventing charring of the tissue around the electrode tip which allows a greater volume of tissue to be treated (before tissue resistance prevents further current flow). The alternating electric current is delivered in a pulsed and cyclic manner and successful ablations usually increase the temperature of the tissue to between 60 C and 80 C. 2.6 Summary HCC and metastasis from colorectal carcinoma are the two most common malignant tumours to affect the liver. The majority of malignant liver tumours remain inoperable due to their number, distribution and/or the presence of residual disease. Therefore, a number of alternative therapies have been developed including hepatic artery infusion as well as non-resectional chemotherapy, PEI, cryoablation, microwave coagulation therapy, laser-induced thermotherapy, and radiofrequency ablation. The theoretical disadvantages of these methods are that they may produce adverse side effects, and none have yet demonstrated a long-term survival advantage. The safety and efficacy of radiofrequency ablation in comparison to other non-resectional (and resectional) therapies needs to be considered. 5

15 3.0 METHODS 3.1 Literature Search Protocol Inclusion Criteria Articles were selected for inclusion in this systematic literature review of radiofrequency ablation on the basis of the following criteria. Participants Human studies on individuals with either primary hepatocellular carcinoma (HCC) or metastatic colorectal liver carcinoma were included in this review. Patients must not have had additional disease treated at the same time nor have had recurrent liver disease. New Intervention Articles concerned with the new intervention,, were included. was defined as: - radiofrequency ablation () of liver tumours conducted with any of the commercially available radiofrequency needle designs. Comparative Interventions The following surgical and non-surgical ablative therapies were included as comparative interventions: Surgical treatment as defined as 1. Surgical resection 2. Hepatic artery infusion chemotherapy (HAIC) Non-surgical ablative treatment modalities as defined as - 1. Percutaneous ethanol injection (PEI) 2. Cryoablation 3. Microwave coagulation therapy (MCT) 4. Laser-induced thermotherapy (LITT) Outcomes Each article included must have contained information on at least one of the following outcomes of the new intervention as compared to the comparative interventions: 1. Perioperative and postoperative mortality of patients 2. Perioperative and postoperative morbidity of patients including either, but not be limited to; - infection - bleeding 6

16 - bile leaks - injury to other structures - discomfort and/or pain 3. Perioperative and postoperative factors for patients including, but not be limited to; - operative time - re-operation/re-intervention - operative/intervention failure rate and/or rate of recurrent or persistent disease (local recurrence adjacent to tumour ablation area) - rate of new disease not confined to the liver - completion of ablation and/or resection 4. Convalescence of patients including, but not be limited to; - length of hospital stay - time until resumption of usual activities (including work) - quality of life measures - postoperative care requirements 5. Evaluation of intraoperative guidance during treatment and follow-up therapeutic response (treatment outcome) imaging involved, but not be limited to; - ultrasound (US) - computed tomography (CT) Types of Studies Randomised controlled trials, quasi-randomised controlled trials and non-randomised comparative studies assessing patients treated with and either one or more other comparative invention/s were included for review. Where appropriate, additional relevant published material in the form of case series and case reports, letters, conference material, editorials and abstracts were retrieved for background information and discussion but otherwise excluded. Language and Publication Date Restriction Searches were conducted without language restriction. There was also no restriction on date of publication. Exclusions The ASERNIP-S Researcher and Protocol Surgeon excluded references that did not meet the inclusion criteria. The reasons were documented when this occurred. 7

17 3.2 Literature Search Strategies Table 1: Databases searched Database Platform Edition MEDLINE Ovid 1966 to Week NLM Gateway Searched from Week 16 to Week PREMEDLINE Ovid Searched from Week 16 to Week PREMEDLINE and MEDLINE Ovid 1966 to Week EMBASE Ovid 1980 to Week Current Contents Ovid 1993 to Week Cochrane Library Issue 3, 2002 Science Citation Index Web of Science Searched April Clinical Trials Database (US) Searched April NHS Centre for Research and Dissemination (UK) C.EXE/nhscrd/newsearch Searched April NHS Health Technology Assessment (UK) Searched April National Research Register Issue 4, 2001 (UK) nrr.htm EORTC Protocols Database Searched April National of Institute Health (US) Searched April CancerLit (US) Searched April Abbreviations: NHS, National Health Service; UK, United Kingdom; US, United States 8

18 Search Terms Search terms used for MEDLINE (Ovid and Ovid Pre-MEDLINE), EMBASE and Current Contents - 1. resection OR surgery 2. cryo* 3. microwav* 4. hepatic artery 5. intraarterial drug administration 6. laser surgery 7. catheter ablation 8. artificial embolism 9. ethanol injection OR alcohol 10. radiofrequency OR radio frequency OR radio-frequency 11. liver carcinoma OR liver cancer OR hepatocellular carcinoma OR liver cell carcinoma OR 2 OR 3 OR 4 OR 5 OR 6 OR 7 OR 8 OR AND AND 13 Search terms used for MEDLINE (NLM Gateway), Science Citation Index, Clinical Trials Database, NHS CRD; NHS HTA and National Research Register - liver carcinoma OR liver cancer OR hepatocellular carcinoma AND (radiofrequency OR radio frequency OR radio-frequency) AND (ethanol OR surgery OR laser OR arterial OR cryo* OR hepatic artery OR microwave ) The ASERNIP-S Researcher used a different, broad strategy for The Cochrane Library database as restricted searches turned up very few references. The simple search term was - radiofrequency OR radio frequency OR radio-frequency Note: * is a truncation character that retrieves all possible suffix variations of the root word e.g. surg* retrieves surgery, surgical, surgeon, etc. In Cochrane the truncation character is *; in Current Contents, EMBASE and MEDLINE (Ovid) it is $. # is a wildcard symbol that substitutes for one required character in Current Contents, EMBASE and MEDLINE (Ovid). Pearling Additionally, the reference list of each article included in the database as a result of the electronic search was hand searched to find other articles not otherwise identified in the electronic searches. 9

19 3.3 Assessment Methods Data from the included studies were extracted by the ASERNIP-S Researcher using standardised data extraction tables developed a priori and checked by a second researcher. Table 1 in Appendix A contains the guidelines 37 that were used to assess the level of evidence of the studies. Data were only extracted if reported in the text, tables, graphs or figures of the article, or could be accurately extrapolated from the data presented. Conversely, if a particular complication was not reported, it was assumed to be unreported rather than not having occurred. For example, if the re-operation rate was not reported in a study, no value was tabulated. This was done to avoid the bias caused by incorrectly assigning a value of zero to an outcome measurement on the basis of an unverified assumption. Safety and efficacy outcomes are tabulated in Appendices C3 to C6. An evaluation of recurrence at follow-up detected by CT imaging is presented in Appendix C7. Critical appraisal of each included study was undertaken with regards to the method of patient selection, comparability of the patient groups, completeness of follow-up, and any other feature of the study design or execution that may have introduced bias (Appendix C2) Data Analysis and Statistics It was not considered appropriate to pool results across studies. Relative risks (RR) or weighted mean differences (WMD) and the 95% confidence intervals (CI) were calculated individually for the same outcomes in the four RCTs 43,47,59,60 and the one quasi-rct. 58 Results were interpreted such that was better than the comparative intervention when the upper limit of the 95% CI of the RR was <1 (except for positive outcomes such as therapeutic effect) and for WMD <0. The RRs, WMDs and 95% CIs were calculated with RevMan 4.1 (Update Software Ltd. 2000) Outcome Measures For this review, the question of safety and efficacy were addressed in terms of whether radiofrequency ablation for the treatment of primary hepatocellular carcinoma (HCC) or metastatic colorectal liver carcinoma was more or less likely to cause harm or injury to the patient in comparison to: Surgical treatment defined as - surgical resection hepatic artery infusion chemotherapy (HAIC) Non-surgical ablative treatments defined as percutaneous ethanol injection (PEI) cryoablation microwave coagulation therapy (MCT) laser-induced thermotherapy (LITT) Therefore, safety outcomes for radiofrequency ablation were assessed in terms of common endpoints reported for surgical treatment or non-surgical ablative treatments, which included; Perioperative and postoperative mortality Perioperative and postoperative morbidity which included; 10

20 - infection; bleeding - bile leaks - injury to other structures - discomfort and/or pain In addition, any other reported outcomes that related to patient safety were also tabulated (Appendix C.3 RCTs and quasi-rcts and Appendix C.5 Non-randomised comparative studies). In terms of efficacy, the question was whether radiofrequency ablation produced at least equivalent clinical outcomes, in comparison to surgical treatment or non-surgical ablative treatments. The indicators chosen to assess the efficacy included; perioperative and postoperative factors for patients operative time re-operation/re-intervention operative/intervention failure rate and/or rate of recurrent or persistent disease rate of new disease not confined to the liver completion of ablation and/or resection Convalescence of patients included; length of hospital stay time until resumption of usual activities (including work) quality of life measures; postoperative care requirements Any other efficacy outcomes also tabulated (Appendix C.4 RCTs and quasi-rcts and Appendix C.6 Non-randomised comparative studies). 4.0 Description of studies A total of 26 comparative studies were identified. Fourteen of these studies were excluded (see Appendix B Table 1). Of the twelve included studies, four were RCTs (one of which was a multi-centre trial )43,47,59,60 one was a quasi-rct 58 and seven were retrospective non-randomised studies 38,40,46,55-57,61 (see Table 2). Their study profiles are given in Appendix C.1. In the quasi- RCT 58, allocation to the different treatments was based on the patient s proximity to the hospital and therefore was not concealed. The seven included nonrandomised studies were likely to have been retrospective comparisons of with previous standard care. 38,40,46,55-57,61 There are a number of clinical trials currently being conducted on the procedure of. To date, the results of these trials have not yet been published and could not be included for review. A list of these ongoing or as yet unpublished studies is given in Appendix B Table 2. 11

21 Table 2: Included studies Authors Intervention Level of Evidence 12 Tumour Type Lencioni et al. compared to II HCC (1999) 43 Olschewski et al. compared to PEI II HCC (2001) 60 Shibata et al. (2002) 59 compared to II HCC MCT Shiina et al. (2000) 47 compared to PEI II HCC Livraghi et al. compared to PEI III-1 HCC (1999) 58 Catalano et al. compared to III-3 HCC (2000) 55 MS-PEI, SS-PEI or LITT Catalano et al. compared to III-3 HCC (2001) 56 MS-PEI, SS-PEI or LITT Izumi et al. (2001) 57 compared to III-3 HCC MCT Kouyama et al. compared to III-3 HCC (2000) 38 MCT Lees and Gilliams (1999) 40 compared to LITT III-3 Colorectal metastatic liver carcinoma Livraghi et al. compared to III-3 HCC (2000) 46 HAIC Yu et al. (2002) 61 compared to surgical resection III-3 HCC Abbreviations:, radiofrequency ablation; PEI, percutaneous ethanol injection; MCT microwave coagulation therapy; MS PEI, multiple session percutaneous ethanol injection; SS PEI, single session percutaneous ethanol injection; LITT, laser-induced thermotherapy; HAIC, hepatic artery infusion chemotherapy Four studies compared and PEI. 43, 47, 58,60 One study compared and LITT. 40 Another study compared and HAIC 46 while three compared and MCT. 38,57,59 One retrospective study compared to surgical resection. 61 Two retrospective comparisons 55,56 by Catalano reviewed patient CT scans for tumour recurrence and nodular changes after treatment with either, multi-session PEI (MS PEI), single-session ( one-shot ) PEI (SS PEI), or LITT for HCC. It is unclear whether the later article 56 is a partial subgroup analysis including some of the patients from the study population of the earlier report. 55 The LITT results from these two articles are reported in the compared to LITT section. The Catalano studies 55,56 also conducted an evaluation of recurrence at follow-up detected by CT imaging. The results from this evaluation are discussed in the results (section 5.6). Evaluation of follow-up therapeutic response was performed via dynamic CT imaging 59,60 or dual phase (arterial phase and portal venous phase) spiral CT imaging. 43,47,57,59 In one study all patients were evaluated by unenhanced CT and additionally by either dual phase spiral CT, triple phase spiral CT or dual phase plus targeted delay spiral CT. 55 Combined assessment by unenhanced CT and dual phase spiral CT was reported in another study. 56 Kouyama et al. (2000) 38 reported the use of enhanced CT. Three studies did not comment on the type imaging that was used for follow-up evaluation. 40,46,61 12

22 4.1 Assessment of Study Methodology Participants were not blinded to the procedure they were undergoing in any of the included 38,40,43,46, 47,55-61 studies and assessor blinding was either not stated or did not occur. Follow-up time was rarely stated, with the longest clinical follow- up time being 44 months postoperatively. 55 The basis of patient selection was stated in four of the twelve studies ,60 Eight patients were excluded from review in one study 57 and 87 patients were excluded from another 55 - in both studies this was due to the presence of residual viable tumour detected at posttreatment examination. The third study investigated 116 consecutive patients. 56 In the fourth study, 122 patients were not included for review, although a reason for this was not stated. 61 Loss to follow-up was stated in one study where eight patients died during the observation period. 55 In one RCT, patients were allocated to either treatment group by sealed envelope 59 so that patient group assignment was probably concealed to both the investigators and the patients. For the other three RCTs, allocation to either treatment was not stated 40,46,47 The quasi-rct, allocated patients into either treatment group according to their location to the hospital. 62 This form of allocation, although not concealed, is not based on patients request or health status and may reduce the chance of selection bias on part of the investigators and patients. Most of the included studies concentrated on reporting efficacy outcomes such as completeness of tumour ablation, number of sessions, rather than reporting on complications and other safety outcomes. Two studies reported using more than one type of electrode in their studies. 40,58 The postoperative outcomes for either study did not distinguish between the different electrodes thus it was unclear whether there were differences with respect to patient outcomes for the different electrode types Units of Analysis The units of measurement for the reported outcomes varied between the studies. Outcomes were reported for the proportion of nodules treated in each group or the number of treatment sessions in each group. Some studies referred to the proportion of patients with a particular outcome; this was mainly for pain, analgesic requirements or complication rates. Eleven of the studies stated the number of patients in each treatment group. 38,43,46,47,55-61 Lees and Gillams 40 reported the number of nodules in each treatment group. 5.0 RESULTS 5.1 Radiofrequency ablation () compared to percutaneous ethanol injection (PEI) Six studies (three RCTs, 43,47,60 1 quasi-rct 58 and two retrospective comparisons 55,56 ) compared PEI with. The level II evidence 43,47,60 was limited by small sample size, short follow-up period and lack of blinding of study participants. Since no outcomes were comparable between these three RCTs, no pooling of data was possible. In the quasi-rct, one application of was used for each lesion (one treatment cycle) while PEI was delivered as multiple sessions (MS PEI) per lesion (one treatment cycle). 58 One RCT 60 stated that one application of was used for each lesion but did not state how many sessions of PEI were performed per lesion (sessions per one treatment cycle). The other two RCTs did not indicate the number of 13

23 applications per lesion nor how many sessions of PEI were performed per lesion (sessions per one treatment cycle). 43,47 The two retrospective comparisons (level III-3 evidence) by Catalano were observational and retrospective in nature and reviewed patient CT scans for tumour recurrence and nodular changes after treatment with either, MS PEI, single-session ( one-shot ) PEI (SS PEI) or LITT for HCC. 56 It is unclear whether the later article 56 is a partial subgroup analysis including some of the patients from the study population of the earlier rep5the LITT results from these two articles are reported in the compared to LITT section. Efficacy RCTs and quasi-rcts Sessions required In one RCT, 43 fewer sessions were required to achieve complete tumour ablation in the group compared to the number required in the PEI treatment group (p<0.01) although it was unclear if this was per patient or per nodule treated. In another RCT, 47 the group also required less sessions (p<0.0001), although it was not clear whether sessions were achieving complete ablation. In the quasi-rct fewer treatment sessions were required per nodule when patients received compared to PEI. 58 A complete ablative response was present in 90% of treated nodules and in 80% of PEI treated nodules (p=0.127). 58 Therapeutic response (defined by CT at 4 months) and long-term tumour control In one RCT 43 and in the quasi-rct, 59 no statistically significant differences were seen in therapeutic response between and PEI as measured by the percentage of nodules showing complete necrosis. In another RCT, no difference could be detected after a single session (91% of treated compared to 82% of PEI treated nodules after 1 cycle of PEI (RR 1.11, 95% CI 0.98 to 1.26). 60 Recurrence In one RCT, 43 there was a trend to less local tumour recurrence (after a median of 14 months) for treated tumours than for PEI treated tumours (in those tumours considered to have undergone complete ablative responses) although this did not quite reach statistical significance (RR 0.24, 95% CI 0.05 to 1.04). In the same RCT, 43 long-term tumour control (mean 16 months) was greater for than PEI treated tumours (RR 1.23, 95% CI 1.02 to 1.50). In another RCT 47, no local recurrence of tumours was reported for the group at 4 months follow-up, compared to 3.5% for the PEI group (RR 0.31, 95% CI 0.01 to 7.38). New lesions were reported to have formed in 10% of treated patients and in 14% in PEI treated patients (RR 0.7, 95% CI 0.17 to 2.87). 47 Survival/Mortality In one RCT, local recurrence-free survival at one year and two years was 98% and 96% in the treatment group compared to the PEI treatment group (83% and 62%) (p=0.0016). 60 At 14

24 one and two years, event-free survival was 86% and 64% respectively in the treatment group compared to the PEI treatment group (77% and 43%) (p=0.012). 60 In one RCT, mortality at one year was 0% in the treatment group compared to 4% for the PEI treatment group (RR 0.19, 95% CI 0.01 to 3.91). 60 At 2 years mortality was 2% in the treated group and 12% in the PEI treated group (RR 0.16, 95% CI 0.02 to 1.28). 60 (These figures have not been adjusted for major prognostic factors, since these factors were not reported.) Operating time and length of hospital stay In the quasi-rct, 58 procedure time was shorter for PEI than for. In one RCT, patients undergoing spent less days in hospital compared to PEI treated patients (p<0.0001). 47 Pain In one RCT, 47 analgesia was required more often with treatment as compared to PEI treatment (RR 3.13, 95% CI 1.90 to 5.14). In the quasi-rct, 58 five patients undergoing had their treatment interrupted due to severe pain. However it was not stated whether this was the case for any PEI patients. One patient in the PEI group required posttreatment analgesia, while two patients in the group were in pain for three to four days after surgery and required analgesics. Retrospective comparisons In one study, a larger proportion of nodules showed an increase in diameter from baseline after treatment with either (58%) or SS PEI (71%) compared to nodules treated by MS PEI (29%). 55 Nodules also appeared to be more round or oblong in shape when treated by either or SS PEI than nodules treated by MS PEI. 55 Residual viable tumour tissue (which tended to correlate with irregular nodule borders) was present in 35% of treated nodules, 54% of MS PEI treated nodules, 68% of SS PEI treated nodules. 55 In the second study (which may have had some patients in common with Catalano et al. (2000) 55, patients who did not show residual viable tumour on the first posttreatment CT scan but developed tumour recurrence (either local or non-local) on follow-up CT scan (3-22 months after the last treatment session) were evaluated with imaging. 56 Nodules with local recurrence (viable tumour tissue within or around the nodule edge but in continuity with the edge of the treated nodule) were associated more often with either (16%) or MS-PEI (14%) treatment than SS-PEI (9%) treated nodules. 56 New non-local nodule formation (either within the same or within different liver segments from the treated nodule) was more common after MS-PEI (62 or (23 treatment than after SS-PEI (14 treatment. 56 Due to difficulties in interpreting the reported data, it was not possible to associate new non-local nodule formation with the number of nodules with local recurrence. In addition, the study did not report which patients showed development of new nodules. Safety In one RCT, high fever for three or more days was reported in 28% of the sessions and 10% of the PEI sessions performed (RR 2.80, 95% CI 1.59 to 4.92)

25 No other complications were reported in any of the patients. 43,47 In the third RCT, 60 56% of treated patients and 66% of PEI treated patients were without serious side-effects or complications (RR 0.76, 95% CI 0.54 to 1.06). In the quasi-rct, 58 2% (heamothorax) of patients developed major complications within 24 hours of treatment and 8% developed minor complications (intraperitoneal bleeding, heamobilia, pleural effusion, mild cholecystitis). Transaminase levels were also 2-4 times higher than the baseline measurements in all patients. The complication rates or transaminase levels for PEI patients were not given. In the retrospective study by Catalano et al. (2000) 55, an inflammatory granulation rim representing non-viable tumour (necrotic) tissue was seen mainly after or MS PEI. This rim was detected as a margin of slightly hyperattenuating tissue on arterial, portal or delayed CT phase. Complications were most common in the SS PEI treatment group (e.g. biliary duct damage in 21%, local atrophy of the liver surface in 29%, and perihepatic effusion in 29% of SS PEI treated patients) than in the or MS PEI groups, where individual complication rates generally occurred in less than 10% of patients. 55 No safety outcomes were reported in the second Catalano 56 study. 5.2 Radiofrequency ablation () compared to cryoablation There were no included studies that compared these two procedures. 5.3 Radiofrequency ablation () compared to microwave coagulation therapy (MCT) Three studies compared MCT and, one RCT (level II) 59 and two non-randomised comparative and possibly historically controlled studies (level III-3). 38,57 The RCT was limited by a lack of patient blinding and small sample size with a short follow-up period. 59 For the study by Izumi et al. (2001) 57, only the outcomes for recurrence could be included, since the results for all the other outcomes were not reported separately for each treatment group. In addition there were only 16 patients in the group, compared to 68 in the MCT. 57 Efficacy In the RCT, the mean number of treatment sessions per nodule was less for compared to MCT (WMD -1.3 sessions, 95% CI to -0.94). 59 Only 10% of nodules required more than one treatment session while 76% of nodules required more than one MCT treatment session, the majority of nodules requiring three MCT treatments. 59 The mean procedure time that was required per session was higher for than for the MCT procedure. 59 Complete therapeutic effect was seen in 96% of treated nodules and 89% of MCT treated nodules (RR 1.08, 95% CI 0.96 to 1.21). 59 During follow-up (mean 18 months), 8% of treated nodules compared to 17% of MCT treated nodules showed residual foci of disease (RR 0.48, 95% CI 0.14 to 1.48). 59 Half of these MCT treated nodules showed an incomplete treatment effect after MCT treatment but no value was stated for the group. The historically controlled study by Izumi et al. (2001) 57 indicated that after a median followup of 18 months (range 7-39 months), 87% of treated patients and 70% of MCT treated patients had not developed new HCC nodules elsewhere in the liver, although this was not statistically significant (chi-square test with Yates correction to compare proportions between groups). 16