Colorectal cancer (CRC), the third most common cancer globally, presents a growing challenge with its increasing disease burden (1). Approximately 20-30% of CRC patients are diagnosed with synchronous liver metastases (CRLM), the most prevalent form of liver malignancy (2). Moreover, around 40% will experience the onset of metachronous liver metastases as the disease advances (3). Surgical liver resection (LR) remains the gold standard in treating CRLM (4). However, its applicability is often limited by operational challenges, patient performance status, and comorbid conditions (4, 5). In such contexts, radiofrequency ablation (RFA) has emerged as a promising technique in the treatment of patients with colorectal liver metastasis (6). RFA presents as an efficacious alternative, offering targeted ablation of metastatic lesions with the advantages of a minimally invasive approach and reduced complication rates (7). Its effectiveness is predominantly observed in cases with limited metastatic burden, optimally in patients harboring a solitary metastasis or multiple metastases, each confined to a size of no more than 3 cm (8). Nonetheless, integrating RFA with surgical resection is recommended to achieve a complete tumor clearance (R0 resection), or as a liver-sparing strategy in scenarios where surgical resection is challenged by difficult tumor locations (9). For instance, this combined approach can improve local control of the tumor in patients with multiple liver lesions (≥4), or in cases where tumors are located near major blood vessels (10). In the latter case, the ‘heat sink effect’ - where flowing blood absorbs and dissipates heat, preventing the tissue from reaching the necessary temperature for complete tumor destruction - can reduce the efficacy of RFA alone, making its combination with surgical resection a more effective strategy (11). However, the use of RFA can be constrained when treating excessively large lesions or those near large blood vessels (9). Additionally, the long-term outcomes of RFA compared to surgery are yet to be established, as evidenced by the lack of published randomized controlled trials. These limitations may have restricted the broader clinical application of RFA.

The aim of the present umbrella review was to systematically evaluate and synthesize the existing evidence from multiple systematic reviews and meta-analyses comparing the role of RFA to surgical resection in the treatment of CRLM.

We performed an umbrella review, a comprehensive and systematic evaluation of multiple systematic reviews and meta-analyses, to assess a broad range of postoperative outcomes by synthesizing data on various surgical treatments in CRLM patients (12). This umbrella review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance (13). Review protocol was registered with the PROSPERO international prospective register of systematic reviews (CRD42024497886).

Surgical resection is the preferred initial treatment for CRLM, but it may not be feasible in cases of challenging anatomical locations or poor patient health (5). RFA is a minimally invasive alternative that has emerged as a recognized and viable treatment option for small-sized CRC liver metastases (27–29). This technique employs a needle that administers an alternating current to produce high heat for tumor destruction, now enhanced by innovative devices featuring expandable electrodes or internally cooled tips (30–32). Despite its increasing utilization in clinical practice, certain factors can restrict its adoption including large or numerous lesions, or those situated near large blood vessels or the surface of the liver, which can compromise the effectiveness of the procedure and heighten the likelihood of complications (33, 34). Moreover, the long-term efficacy and safety of RFA in the treatment of CRLM compared to surgical resection remain a subject of ongoing debate within the literature. Most of the existing evidence on the efficacy of RFA for CRC liver metastases is derived from retrospective studies. Many of these studies are characterized by limited follow-up periods, often not exceeding 20 months, posing challenges in conclusively determining the long-term outcomes of RFA treatment. The results of our umbrella meta-analyses provide valuable insights into the comparative efficacy of RFA and LR in treating hepatic metastases of CRC, with data encompassing up to a long-term period of 5 years.

A significant survival benefit was observed in patients undergoing LR compared to those who had RFA. This was evident in both short-term (3-year OS) and long-term (5-year OS) follow-ups. The superiority of LR was also pronounced in patients with solitary hepatic lesions. These results suggest that RFA, although being a less invasive technique, might not offer a survival benefit. When comparing a combination of LR and RFA (LR+RFA), a survival advantage was observed in the group undergoing only resection. While this advantage was not always statistically significant, it suggests that RFA may not play a significant role in enhancing survival.

The analysis of DFS demonstrates a clear advantage for LR. This benefit is evident when comparing DFS outcomes of LR against implementation of RFA, whether the latter is used alone or in conjunction with surgery. This is particularly notable in the 5-year DFS comparisons, where LR alone shows a statistically significant benefit. The distinction is less pronounced when comparing LR to a combination treatment with RFA, yet it still indicates a trend towards improved outcomes.

Regarding perioperative mortality, our data did not reveal significant differences between RFA and surgical treatments. However, what stands out is the markedly lower rate of postoperative complications observed in the RFA group. This trend persists even when RFA is used in conjunction with surgery, suggesting a consistently reduced risk of complications. This aspect of our findings is particularly significant, as it suggests that RFA, either alone or in combination with surgery, might offer a safer treatment path with potentially less intensive postoperative care requirements. This lower complication profile could be a decisive factor in treatment choice, especially in patients where minimization of postoperative risk is a primary concern. However, a critical finding was the substantially higher rate of intrahepatic recurrence in patients treated with RFA. This suggests that while RFA may offer better initial outcomes in terms of perioperative complications, there is a higher risk of recurrence, which could impact long-term survival and quality of life. As RFA resembles the treatment effect of a wedge (i.e., non-anatomical) liver resection, this result confirms the previous evidence that the anatomical removal of the portal tributaries of a given metastasis reduces the risk of liver recurrence after surgery (35, 36). Notably, the risk of recurrence may be influenced by several prognostic factors, including bilobar disease, positive surgical margins, and elevated tumor markers (37). Additionally, molecular markers such as RAS, BRAF, and SMAD4 mutations significantly impact overall and recurrence-free survival, particularly in patients undergoing liver metastasectomy (38). These genetic factors may explain the higher intrahepatic recurrence rates observed after RFA, reinforcing the importance of careful patient selection. At the same time, technical factors remain crucial. Studies have shown that achieving an ablative margin greater than 5 mm during RFA significantly reduces local tumor progression (39). Advanced imaging techniques, particularly contrast-enhanced MRI, can improve treatment planning by providing a more accurate assessment of tumor extent and guiding surgical or ablative approaches (40). Ensuring precise margins, whether through surgery or locoregional therapies, is key to minimizing recurrence risk. Ultimately, optimizing CRLM treatment requires a multidisciplinary approach. The decision between surgery, RFA, or other locoregional therapies should balance both technical feasibility and tumor biology. While advancements in imaging and ablation techniques have improved outcomes, their success still depends on careful planning and patient selection.

A critical limitation of our study, and indeed a potential confounding factor in our results, stems from the selection criteria for the patient cohort undergoing RFA. Specifically, these patients were those considered ineligible for surgical resection, and that inherently places them at a survival disadvantage relative to candidates for LR. This aspect suggests that the observed differences in treatment efficacy might not solely reflect the intrinsic merits of the treatment modalities but could also be influenced by the baseline survival prospects of the patient groups involved. Other limitations must be acknowledged as they potentially impact the interpretation of our umbrella review results. The quality of the included meta-analyses, assessed using AMSTAR-2 criteria, was found to be generally critically low, necessitating prudence in the interpretation of their findings. Future studies should aim to mitigate the impact of selection bias arising from patient and tumor characteristics in surgical treatment decisions through rigorous sensitivity analyses while ensuring methodological robustness by adhering to AMSTAR-2 criteria, thereby strengthening the validity and reliability of evidence in surgical oncology research. Additionally, the studies predominantly comprise non-randomized controlled trial data, including retrospective cohort studies, which could potentially introduce bias. To mitigate these issues, we performed a survival analysis using only matched cohort analysis studies, pairing patients who underwent RFA alone with those who underwent LR alone, based on prognostic characteristics. This approach was designed to simulate the randomization process typical of clinical trials, aiming to minimize selection bias and thereby enhance the reliability of our conclusions. While the analysis of disease-free survival revealed a benefit for surgical resection over RFA, both treatment modalities yielded comparable outcomes in terms of overall survival.

Our study represents the first umbrella review to compare RFA and LR in the management of CRLM, offering a clear overview of the existing evidence on this topic. As suggested by our results, LR maintains its crucial role in the management of CRLM, given its advantages in both survival and disease-free survival. However, given its significantly lower rate of postoperative complications, RFA stands out as a less invasive treatment approach in CRLM management in selected comorbid patients. Nevertheless, to fully ascertain its impact on survival and quality of life, further long-term follow-up studies and randomized controlled trials are needed. Future investigations should: (1) focus on specific patient subgroups, such as those with solitary CRLM ≤3 cm, to clarify the efficacy of RFA in these scenarios; (2) compare patients with similar baseline characteristics in terms of prognosis to minimize selection bias; (3) employ long-term follow-up studies (beyond 5 years) while investigating causes of higher recurrence rates with RFA. Future systematic reviews and meta-analyses should include more high-quality randomized controlled trials, such as the ongoing COLLISION trial, to compare RFA and LR. Such research would provide a clearer and more definitive comparison of the efficacy of LR versus RFA, thereby offering more reliable guidance for clinical decision-making.