Over the past 30 years, several ablative methods have been developed for the treatment of hepatic cancer as an alternative to surgical resection and liver transplantation in patients with unresectable cancer or in selected patients with resectable disease. Recent guidelines recommend radiofrequency ablation (RFA) and microwave ablation (MWA) as the ablative methods with the highest efficacy in the treatment of liver cancer.1, 2
Tumor location near to the main biliary tree, abdominal organs, or diaphragm is a relative contraindication for RFA because of the risk of severe complications.2 RFA is prone to heat-sink effect, which reduces further the efficacy of the treatment.1, 2 On the other hand, MWA, which is a more recent thermal ablation technique, is associated with higher intratumoral temperatures, resulting in faster, larger, and more homogenous ablation compared to RFA.2 Furthermore, MWA is less prone to heat-sink effect and can be utilized in tumors adjacent to vessels.
According to the European Association for the Study of the Liver (EASL) recommendations, MWA showed promising results for local tumor control in patients with hepatocellular cancer (HCC).1 The guidelines by the American Association for the Study of Liver Diseases (AASLD) reported that MWA has potential advantages over RFA; however, further studies are required to provide safety and efficacy data.2 The Cochrane meta-analysis conducted in 2013 failed to provide evidence regarding the role of ablative methods in the treatment of HCC since only one randomized clinical trial (RCT) with high risk of bias was available.3 The last conducted meta-analysis in 2019 reported beneficial outcomes in favor of MWA.4 However, low quality randomized and observational studies, which were affected by confounding bias were included in this meta-analysis, which could influence the reliability of the outcomes.
Despite the promising results of MWA in the treatment of liver cancer, efficacy and safety of MWA compared to RFA is unclear. Aim of this meta-analysis is to compare RFA and MWA in the treatment of HCC and liver metastases. Our hypothesis is that the beneficial characteristics of MWA are translated into better oncological outcomes compared to RFA.
A protocol was developed to pre-specify criteria for including and excluding studies in the review. Eligibility criteria were based on the PICO elements (population, interventions, comparators, and outcomes) plus a specification of the type of studies that have addressed these questions. RCTs and observational studies (prospective or retrospective cohort and case-control studies) were eligible for inclusion. Studies conducted before 2010 were excluded from the meta-analysis.
Studies meeting the following criteria were included: (1) population: adults with primary liver cancer or hepatic metastases; (2) interventions: RFA and MWA as monotherapy or combined with surgical resection; (3) MWA and RFA conducted percutaneously, laparoscopically, or through laparotomy; (4) comparators: effectiveness and safety of MWA compared to RFA; (5) outcomes: results provided data relative to complete ablation (CA), local tumor progression (LTP), intrahepatic distant (IDR), complications; (6) full text available in English or German; (7) studies with low or moderate risk of bias.
Exclusion criteria were the following: (1) studies with benign liver tumors; (2) pediatric population; (3) animal or in vitro studies; (4) RFA or MWA combined with other interventions such as transarterial chemoembolization (TACE); (5) gender and geographical criteria were not utilized; (6) stage of liver cancer, size, and location of tumors did not constitute exclusion criteria; (7) duplicate data.
Primary outcomes were the CA rates and the LTP. CA was defined as the no enhancement of the tumor in the hepatic arterial or portal venous phase in dynamic enhanced imaging (CT, MRI), which was conducted after ablation. As incomplete ablation was defined the enhancement of the tumor in dynamic enhanced imaging.5 As LTP was defined the reappearance of the tumor within or adjacent to the ablation zone during the follow-up period. Studies that reported recurrence rates without to define if that is local or distant were excluded from this analysis. In studies where 1-year, 3-year, and 5-year LTP rates were reported, only the overall 5-year LTP rate was included in the analysis. In the majority of cases, patients were presented with multinodular disease. For that reason, CA and LTP were recorded for every treated lesion. Studies, where LTP and CA were recorded per patient and not per lesion, were excluded from the analysis.
IDR and complications were included in the secondary outcomes. IDR was defined as distant recurrence within the liver. In studies where 1-year, 3-year, and 5-year IDR rates were reported, the overall 5-year rate was included in the meta-analysis. Minor complications, which required no intervention or were not associated with prolonged hospital stay, were not included in the analysis. Major complications were defined as post-interventional events that lead to substantial morbidity or disability, require intervention, and result in prolonged hospital stay.
A subgroup analysis was conducted, comparing CA and LTP for tumors ≤ 3 cm and tumors > 3 cm in diameter. RFA and MWA were compared separately in patients diagnosed with HCC and colorectal liver metastases (CRLM).
The systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.6 A systematic search of MEDLINE (PubMed and Ovid) and the Cochrane Central Register of Controlled Trials was conducted for relevant systematic reviews, RCTs, and observational studies. Access to Embase was not available for the review team. The search was accomplished in July 2020.
The search strategy included the following keywords: ((((“Carcinoma, Hepatocellular”[Mesh])) OR (hepatic tumor)) AND (“Radiofrequency Ablation”[Mesh])) AND (“Microwaves”[Mesh]). The search strategy was not limited by geographical criteria. English and German language articles were reviewed for inclusion. Studies conducted between 2010 and 2020 were screened. Reference lists of retrieved studies and relevant reviews were hand-searched.
Eligibility for inclusion was evaluated in the title and abstract of each publication. If the title and abstract were relevant to the review question, full-text screening was conducted. Reviewers were not blinded to the name of authors and institutions. Screening of articles was conducted by two reviewers. Discrepancies were resolved by consensus. If consensus was not reached, discrepancies were resolved by adjudication from a third reviewer. Data were extracted independently by two reviewers and checked from a third reviewer. When further information was required during data extraction, the reviewers tried to contact the corresponding author with email.
Non-randomized studies were included since available RCTs were limited. The quality of RCTs and observational studies was assessed using the Cochrane Risk of Bias version 2 (RoB 2) tool and The Risk of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool, respectively.7 Risk of bias was assessed independently by two reviewers. In case of disagreement, a third author adjudicated the final judgement. High risk RCTs were excluded from the analysis.
Non-randomized studies vary with respect to their intrinsic ability to estimate the causal effect of an intervention. Therefore, to reach reliable conclusions and to eliminate the risk of bias in our results, only studies with low and moderate risk of bias were included in the meta-analysis. Studies with “Serious”, “Critical” risk of bias, or “No information” were excluded from the meta-analysis.
Review authors have defined confounding domains in the review protocol. Confounding domain is a preintervention prognostic factor of the outcome that also predicts whether an individual receives RFA or MWA. Non-randomized studies were assessed as ‘Low Risk of Bias’ in this domain when patients in both groups were matched using propensity score based on the confounding factors. Surveys that compared confounding factors at baseline without propensity score matching and reported no statistical differences were included as studies with ‘Moderate Risk of Bias’. Finally, studies with statistically different baseline characteristics or not reported or not compared baseline characteristics were assessed as ‘Serious Risk’ or ‘Critical Risk’ and were excluded from the analysis.
For all outcomes of interest, meta-analyses for the Odds Ratio (OR) have been performed. The amount of heterogeneity (measured by
The amount of heterogeneity among studies has been measured by the
A total of 716 publications were identified from database searching. After removing duplicates, 581 unique articles were screened for inclusion. During the title-abstract screening phase, a total of 531 irrelevant studies were excluded. Fifty articles were selected for full-text review. Thirty-five articles were excluded because of no comparison between RFA and MWA (n = 17), increased risk of bias in confounding domain for observation studies (n = 14), combined treatment with TACE (n =1), and no relevant outcomes (n = 3). The RCT by Yu
Prisma flow diagram.
MWA = microwave ablation, RFA = radiofrequency ablation, TACE = transarterial chemoembolization
Fifteen studies with a total of 2,169 patients were included in the analysis.14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28 The recruitment period ranged from 2001 to 2018. The sample size ranged from 40 to 460 patients. The average age across studies ranged from 52 to 68 years. The mean or median tumor size ranged from 1.7 cm to 3.75 cm. All studies reported no significant differences in tumor size between the two treatment groups. Study characteristics and baseline characteristics are demonstrated in Table 1.
Study and baseline characteristics of studies included in the meta-analysis
Study | County | Study Design | Tumor | RFA, n | MWA, n | Age, RFA | Age, MWA | Child-Pugh A/B/C, RFA | Child-Pugh A/B/C, MWA | Tumor size (cm), RFA | Tumor size (cm), MWA | Tumor lesions, RFA | Tumor lesions, MWA |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Kamal 2019 (13) | Egypt | RCT | HCC | 28 | 28 | 55 | 55 | 22.6.2000 | 22.6.2000 | 3.28 ± 0.91 | 3.25 ± 0.92 | 34 | 34 |
Vietti Violi 2018 (14) | France/Switzerland | RCT | HCC | 73 | 71 | 65 (median) | 68 (median) | 53/20/0 | 57/14/0 | 1.8 ± 0.71 | 1.8 ± 0.65 | 104 | 98 |
Abdelaziz 2014 (15) | Egypt | RCT | HCC | 45 | 66 | 56.8 ± 7.3 | 53.6 ± 5 | 24/21/0 | 25/41/0 | 2.95 ± 1.03 | 2.9 ± 0.97 | 52 | 76 |
Di Vece 2013 (16) | Italy | RCT | HCC/Metastases | 20 | 20 | 59 (median) | 63 (median) | N/R | N/R | 3.2 (median) | 3.6 (median) | 20 | 20 |
Qian 2012 (17) | China | Prospective | HCC | 20 | 22 | 56 ± 11 | 52 ± 12 | 20/0/0 | 22/0/0 | 2 ± 0.5 | 2.1 ± 0.4 | 20 | 22 |
Sparchez 2019 (18) | Romania | Retrospective | Metastases | 44 | 17 | 60.18 ± 9.96 | 62.12 ± 10.73 | - | - | 2.maj | feb.55 | 62 | 20 |
Takahashi 2018 (19) | USA | Retrospective | CRLM | 54 | 51 | N/R | N/R | - | - | 2.4 (median) | 2.1 (median) | 155 | 121 |
Shady 2018 (20) | USA | Retrospective | CRLM | 62 | 48 | N/R | N/R | - | - | 1.8 (median) | 1.7 (median) | 85 | 60 |
Xu 2017 (21) | China | Retrospective | HCC | 159 | 301 | 54 ± 11 | 54.2 ± 11 | 140/19/0 | 278/23/0 | 1.7 ± 0.3 | 1.7 ± 0.3 | 159 | 301 |
van Tilborg 2016 (38) | Netherlands | Retrospective | CRLM | Total number of participants: 122 | N/R | N/R | - | - | 2.apr | 2.maj | 151 | 48 | |
Potretzke 2016 (23) | USA | Retrospective | HCC | 55 | 99 | 62 | 61 | N/R | N/R | 2.apr | 2.feb | 69 | 136 |
Zhang X. 2014 (24) | China | Retrospective | HCC/Metastases | 92 | 230 | 51.5 ± 14.3 | 55.7 ± 13.2 | N/R | N/R | 5.4 ± 1.9 | 5.7 ± 2.1 | 173 | 349 |
Zhang L. 2013 (25) | China | Retrospective | HCC | 78 | 77 | 54 ± 10.5 | 54 ± 9.5 | 78/0/0 | 77/0/0 | 2.3 ± 0.4 | 2.2 ± 0.4 | 97 | 105 |
Liu 2013 (26) | China | Retrospective | Metastases | 54 | 35 | 53.1 ± 12.7 | 53.4 ± 15.3 | - | - | 2.5 ± 1.0 | 2.3 ± 1.0 | 70 | 62 |
Ding 2013 (27) | China | Retrospective | HCC | 85 | 113 | 58.64 ± 8.52 | 59.06 ± 11.68 | 49/36/0 | 75/30/0 | 2.38 ± 0.81 | 2.55 ± 0.89 | 98 | 131 |
Age and tumor size are recorded as mean, mean ± standard deviation (SD), or median.
CRLM = colorectal liver metastases, HCC = hepatocellular cancer, MWA = microwave ablation, RCT = randomized clinical trial, RFA = radiofrequency ablation, N/R = not reported
Eight studies evaluated the role of thermal ablation in patients with HCC.14, 15, 16,18,22,24,26,28 Child-Pugh score, which was estimated in the majority of studies, was not statistically different between RFA and MWA groups. In the retrospective study conducted by Potretzke
The quality of included RCTs was acceptable (Supporting Information, Figure S1). Two out of four RCTs were judged to be at low risk of bias across all domains.15,17 The RCT conducted by Abdelaziz
Three studies reported the method of randomization and allocation sequence generation. Coin flip16 and centralized computer-generated randomization15,17 were utilized as methods for random sequence generation. In these RCTs, the allocation sequence was adequately concealed. The study by Kamal
Physicians, who conducted the ablations, were not blinded, since different equipment was utilized in each treatment modality. Patients were masked to the treatment in one trial.15 In two RCTs, independent outcome assessors, who were masked to the treatment allocation, reviewed all images and recorded the outcomes.15,17 In the studies conducted by Kamal
All retrospective studies were judged to be at moderate risk of overall and confounding bias (Supporting Information, Table S1). Studies that evaluated the role of ablation in hepatic metastases did not report the histological stage of the primary tumors.19, 20, 21,23,25, 27 Two studies that included HCC patients did not compare the BCLC stage at baseline.25,26 Tumor size was comparable between the two groups in all studies.
Four studies were affected by selection bias.18,20,22, 23 In these studies, the number of excluded patients and the reason of exclusion were not reported. Bias due to deviations from intended interventions was seen only in the survey by van Tilborg
Non-significant difference was found in CA rates between MWA and RFA (OR, 1.10; 95% CI, 0.78– 1.55;
Forest plot of random-effects meta-analysis results for complete ablation rates in the MWA and RFA group.
CI = confidence interval, MWA = microwave ablation, OR = odds ratio, RFA = radiofrequency ablation
LTP rates were comparable between MWA and RFA (OR, 0.79; 95% CI, 0.53–1.20;
Forest plot of random-effects meta-analysis results for local tumor progression in the RFA and MWA group.
CI = confidence interval, MWA = microwave ablation, OR = odds ratio, RFA = radiofrequency ablation
Forest plot of random-effects meta-analysis results for local tumor progression in the RFA and MWA group based on the study type.
CI = confidence interval, MWA = microwave ablation, OR = odds ratio, RFA = radiofrequency ablation
Analysis of seven studies showed no statistically significant differences in IDR between MWA and RFA (OR, 0.73; 95% CI, 0.45–1.16;
Forest plot of random-effects meta-analysis results for intrahepatic distant recurrence rates in the RFA and MWA group.
CI = confidence interval, MWA = microwave ablation, OR = odds ratio, RFA = radiofrequency ablation
The most commonly reported major complications in both groups were subcapsular hepatic hematoma, perihepatic hematoma, arterial bleeding requiring embolization or surgical treatment, hepatic abscess, biliary fistula, bowel perforation, abdominal wall skin burn, and pleural effusion. The risk of major complications was not different between the two approaches (OR, 0.80; 95% CI, 0.46–1.37;
Forest plot of random-effects meta-analysis results for complication rates following RFA and MWA.
CI = confidence interval, MWA = microwave ablation, OR = odds ratio, RFA = radiofrequency ablation
Four studies assessed the rates of CA in patients with tumor < 3 cm.6,18,24,28 Heterogeneity among the surveys was not significant (
Regarding tumors with size larger than 3 cm, three studies reported CA rates16,24,28 and two studies evaluated LTP.24,28 Meta-analysis showed no significant difference in CA and LTP between RFA and MWA (
Meta-analysis showed no significant difference in CA between RFA and MWA in patients with HCC (OR, 1.18; 95% CI, 0.70–1.99;
LTP was not significantly different between RFA and MWA (OR, 0.77; 95% CI, 0.49–1.22,
Differences between RFA and MWA in the incidence of IDR were not found (OR, 0.75; 95% CI, 0.43–1.30;
CA and LTP were compared between RFA and MWA in patients with CRLM. Meta-analysis included three retrospective studies.20,21,23 For both outcomes, no significant differences were found between the two procedures (
CA, LTP, IDR, complications, CA in HCC patients, LTP in HCC patients, and IDR in HCC patients were examined for publication bias (Supporting Information, Figure S3, S4). Results demonstrated a low risk of publication bias for the outcomes assessed. Egger’s test was utilized in the outcomes with more than ten included studies. No obvious asymmetry or
RFA is currently one of the most widely used thermal ablation modalities. On the other hand, utilization of MWA has been increased the last years as a result of significant advancements in technology of new generation devices. These advancements are translated into higher temperatures and faster heating compared to RFA, large ablation volumes, and less heat sink effect.29 However, MWA has not been adequately compared with RFA and selection of appropriate treatment is not based on high level of evidence.30 On the basis of these considerations, we conducted the present meta-analysis to evaluate the role of MWA in the treatment of liver cancer.
Meta-analysis of CA rates, which included more than 2,500 tumor lesions, demonstrated no significant differences between MWA and RFA. In the subgroup analysis of RCTs with 438 tumors, similar rates of CA were found between the two methods. Analysis of all included studies revealed no significant difference in LTP between MWA and
RFA. Since increased heterogeneity was detected among the studies, subgroup analysis of RCTs was conducted to decrease heterogeneity and to evaluate the influence of observational studies on the outcomes. The RCTs by Abdelaziz
The finding of the RCTs is consistent with the physics and characteristics of radiofrequency and microwave energies. MWA is associated with higher temperatures, faster heating, larger ablation volumes, and less heat sink effect compared to RFA, which are translated into better oncological outcomes in terms of LTP in the present meta-analysis. On the other hand, meta-analysis of retrospective studies failed to demonstrate superiority of MWA over RFA, which is attributed to the significant inter-study heterogeneity.
Consequently, though CA was comparable between the two procedures, LTP was beneficial in favor of MWA. These conflicting results are not surprising given the limitations associated with measurement and evaluation of complete ablation response. Imaging modalities cannot detect with 100% accuracy whether neoplastic cells have been sufficiently ablated. For that reason, ablation response cannot be considered as the most reliable indicator of treatment effectiveness. On the other hand, follow-up imaging examinations and LTP have been considered of great importance in detecting treatment failure. LTP is the most reliable indicator of treatment effectiveness and can be utilized as assessment tool of treatment efficacy.
IDR was comparable between the two ablative methods. Subgroup analysis of two RCTs demonstrated similar rates of IDR between MWA and RFA. The RCT by Kamal
The risk of complications was not significantly different between the groups and both procedures presented a limited number of adverse events. This finding is important since larger ablation zones, which are achieved through MWA, could be perceived to cause more perioperative complications and damage to liver function compared to RFA. This assumption was refuted with the results of our meta-analysis.
CA and LTP were compared separately among patients with HCC and CRLM. As mentioned above, results derived from the two RCTs in HCC patients showed statistically decreased rates of LTP following MWA compared to RFA.15, 16 In the present meta-analysis, only three retrospective studies compared the two methods in patients with CRLM; consequently, reliable conclusions cannot be drawn, though results showed no significant difference.
In accordance with our results, previous studies reported similar rates of CA between RFA and MWA.4,31, 32, 33, 34 Glassberg
Subgroup analysis showed no difference between RFA and MWA for tumor size less or larger than 3 cm. Similar to our findings, Luo
In contrast to our results, Glassberg
Contrary to our findings, the meta-analysis conducted by Glassberg
The findings in the present meta-analysis should be interpreted in view of certain limitations. First, observational studies without randomization were included in the analysis, which is associated with potential confounding, selection, measurement, and reporting bias. In order to eliminate bias attributed to observational studies, only surveys with low or moderate overall risk of bias were included. Second, significant inter-study heterogeneity was observed for certain outcomes. In these cases, influence of retrospective studies on the results and sources of heterogeneity were examined with subgroup analysis of RCTs and retrospective studies separately. Third, different MWA and RFA devices were utilized across the surveys, which could influence the results of our analysis. Since various devices were used, a subgroup analysis based on the type of devices was not possible. Fourth, limited number of studies included patients with liver metastases or CRLM. Consequently, further RCTs are required to compare MWA with RFA in patients with hepatic metastases.
In addition, in the present study, the proved superiority of MWA over RFA in terms of LTP cannot be translated into better long-term oncological outcomes, since survival outcomes were not evaluated. Overall survival and disease-free survival were not included in our analysis, since limited data can be drawn from the available studies. The majority of surveys were retrospective in design and have included patients with no 100% matching in oncological characteristics. Furthermore, some patients underwent simultaneously surgical resection and ablation. Survival of these patients is multifactor in etiology and causality. Regarding patients with liver metastases, neoadjuvant or adjuvant treatment and tumor stage were not 100% similar between the two groups. For that reason, survival after ablation is associated with several parameters, which could not be attributed only to the effectiveness of the ablative procedures. In fact, LTP and CA are generally considered the best indicators of treatment effectiveness for ablative methods rather than overall survival or disease-free survival.
The meta-analysis is strengthened by its broad inclusion of 15 studies with a total of 2,169 patients. In contrast to other meta-analyses, low quality studies were excluded. Consequently, results were derived from high or moderate quality studies. Taking into consideration the results of the present meta-analysis, we suggest that MWA should be the ablation method of choice in the treatment of HCC. Finally, since the majority of studies included patients with HCC, further RCTs are required to evaluate the role of ablation treatments in patients with liver metastases.