• Roberto Cirocchi, Stefano Trastulli, Carlo Boselli, Alessandro Montedori, Davide Cavaliere, Amilcare Parisi, Giuseppe Noya, and Iosief Abraha.
• Department of General Surgery, University of Perugia, Terni, Italy. cirocchiroberto@yahoo.it.
• Cochrane Db Syst Rev. 2012 Jan 1;6:CD006317.

BackgroundColorectal cancer (CRC) is the most common malignant tumour and the third leading cause of cancer deaths in USA. For advanced CRC, the liver is the first site of metastatic disease; approximately 50 % of patients with CRC will develop liver metastases either synchronously or metachronously within 2 years after primary diagnosis. Hepatic resection (HR) is the only curative option, but only 15-20% of patients with liver metastases from CRC (CRLMs) are suitable for surgical standard treatment. In patients with unresectable CRLMs downsizing chemotherapy can improve resectability (16%). Modern systemic chemotherapy represents the only significant treatment for unresectable CRLMs. However several loco-regional treatments have been developed: hepatic arterial infusion (HAI), cryosurgical ablation (CSA), radiofrequency ablation (RFA), microwave ablation and selective internal radion treatment (SIRT). During the past decade RFA has superseded other ablative therapies, due to its low morbidity, mortality, safety and patient acceptability.ObjectivesThe objective of this study was to systematically review the role of radiofrequency ablation (RFA) in the treatment of CRLMs.Search MethodsWe performed electronic searches in the following databases:CENTRAL, MEDLINE and EMBASE. Current trials were identified through the Internet using the Clinical-Trials.gov site (to January 2, 2012) and ASCO Proceedings. The reference lists of identified trials were reviewed for additional studies.Selection CriteriaRandomized clinical trials (RCTs), quasi-randomised or controlled clinical trials (CCTs) comparing RFA to any other therapy for CRLMs were included. Observational study designs including comparative cohort studies comparing RFA to another intervention, single arm cohort studies or case control studies have been included if they have: prospectively collected data, ten or more patients; and have a mean or median follow-up time of 24 months. Patients with CRLMs who have no contraindications for RFA. Patients with unresectable extra-hepatic disease were also included.Trials have been considered regardless of language of origin.Data Collection And AnalysisA total of 1144 records were identified through the above electronic searching. We included 18 studies: 10 observational studies, 7 Clinical Controlled Trials (CCTs) and an additional 1 Randomized Clinical Trial (RCT) (abstract) identified by hand searching in the 2010 ASCO Annual Meeting. The most appropriate way of summarizing time-to-event data is to use methods of survival analysis and express the intervention effect as a hazard ratio. In the included studies these outcome are mostly reported as dichotomous data so we should have asked authors research data for each participant and perform Individual Patient Data (IPD) meta-analysis. Given the study design and low quality of included studies we decided to give up and not to summarize these data.Main ResultsSeventeen studies were not randomised and this increases the potential for selection bias. In addition, there was imbalance in the baseline characteristics of the participants included in all studies. All studies were classified as having a elevate risk of bias. The assessment of methodological quality of all non-randomized studies included in meta-analysis performed by the STROBE checklist has allowed us to identify several methodological limits in most of the analysed studies. At present, the information from the single RCT included (Ruers 2010) comes from an abstract of 2010 ASCO Annual Meeting where the allocation concealment was not reported; however in original protocol allocation concealment was adequately reported (EORTC 40004 protocol). The heterogeneity regarding interventions, comparisons and outcomes rendered the data not suitable.Authors' ConclusionsThis systematic review gathers information from several controlled clinical trials and observational studies which are vulnerable to different types of bias. The imbalance between characteristics of patients in the allocated groups appears to be the main concern. Only one randomised clinical trial (published as an abstract), comparing 60 patients receiving RFA plus CT versus 59 patients receiving CT alone, was identified. This study showed that PFS was significantly higher in the group that received RFA. However, it was not able to provide information on overall survival. In conclusion, evidence from the included studies are insufficient to recommend RFA for a radical oncological treatment of CRLMs.

7 keywords...

hide…