A comprehensive search was performed across major databases, including PubMed (Medline), EMBASE, Cochrane Library, and Web of Science, to identify relevant studies published up to December 2024. The search terms included “Microwave ablation” (MWA), “Thermal ablation” (TA), “Radiofrequency ablation” (RFA), “Liver resection” (synonyms: “Hepatic resection,” “Hepatectomy”), “Colorectal” (synonyms: “Colon,” “Rectal”), “Cancer” (synonyms: “Tumor,” “Carcinoma”), and “Liver metastasis” (synonyms: “Hepatic metastases”). Only English-language publications were considered. Additionally, the reference lists of retrieved articles were manually reviewed to identify additional relevant studies.

Studies were included if they met the following criteria: (1) they compared the clinical outcomes of combined thermal ablation (TA) with liver resection (LR) versus LR alone for treating colorectal cancer liver metastases (CRLM); (2) they reported at least 3- or 5-year overall survival (OS) and/or disease-free survival (DFS) for each treatment group; (3) in cases of multiple publications from the same research group, only the most recent and comprehensive study was included; (4) studies involved patients with CRLM (preoperatively or intraoperatively diagnosed with liver metastases); and (5) only randomized controlled trials (RCTs) and non-RCTs published in English were included.

Studies were excluded if they: (1) did not provide comparative data on the therapeutic efficacy of MWA versus HR; (2) were unsuitable publication types (case reports, conference abstracts, meta-analyses, reviews, or animal experiments); (3) included patients with extrahepatic metastases; (4) lacked a control group or had an unreasonable control group; (5) were not written in English; (6) were of low quality; or (7) no original data could be obtained from the corresponding author.

The methodological quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS). Two independent reviewers (Zesong Meng and Baokun Li) evaluated the studies, and any disagreements were resolved through consultation with a third reviewer (Longfei Cao). Studies scoring ≥6 on the NOS were deemed high quality.

Data extraction was independently conducted by Baokun Li and Chaoxi Zhou, with discrepancies resolved through discussion with a third reviewer (Longfei Cao). Extracted data included: (1) study characteristics (first author, publication year, sample size, study location, and design); (2) baseline oncological characteristics; and (3) clinical outcomes (1-, 2-, 3-, and 5-year OS and DFS, as well as perioperative outcomes comparing resection and combined resection and ablation).

For the meta-analysis, when individual patient-level data were available, Kaplan-Meier curves and risk tables were reconstructed. For studies lacking primary data, summary data were extracted from published Kaplan-Meier curves, including treatment group details, overall survival probabilities with corresponding time points, and recurrence-free survival probabilities with associated time intervals (10, 11).

Dichotomous variables were analyzed using odds ratios (OR) with 95% confidence intervals (CI). Inter-study heterogeneity was assessed using I² statistics. Time-to-event data, including 1-, 2-, and 3-year OS, were extracted from individual studies. Pooled categorical comparisons were conducted using the Chi-squared test. A random-effects model (DerSimonian-Laird method) was applied if significant heterogeneity (I² > 50%) was observed; otherwise, a fixed-effect model was used. A two-sided p-value < 0.05 was considered statistically significant. Sensitivity analysis was performed by sequentially excluding each study to assess the stability of the results. Publication bias was evaluated using Begg’s and Egger’s tests. All analyses were conducted using STATA version 12.0.

A total of six retrospective cohort studies, involving 3,084 patients (1,286 in the TA+LR group and 1,798 in the LR group), were included in this meta-analysis ( Figure 1 ). The final analysis encompassed six retrospective cohort studies (11–16). All studies demonstrated high methodological quality, achieving Newcastle-Ottawa Scale (NOS) scores of at least 7. The baseline characteristics of the included studies are detailed in Table 1 . The majority of these studies were conducted in Europe (n=4), while the remaining two studies originated from China and the United States.

A total of 5 publications reported the 1-year OS, 2-year OS and 3-year OS, which was not significantly different between the TA+LR group and LR group at 1-year (OR=0.88, 95% CI: 0.42–1.82; p=0.72; Figure 2A ), 2-year (OR=0.72, 95% CI: 0.49–1.04; p=0.08) ( Figure 2B ), 3-year (OR=0.95, 95% CI: 0.48–1.88; p=0.88) ( Figure 2C ).

A total of 5 publications reported the 1-year DFS, disease-free survival (DFS) did not differ significantly between groups at 1-year (OR=0.78, 95% CI: 0.45–1.33; p=0.36, Figure 3A ). Patients in the TA+LR group had significantly shorter 2-year DFS (OR: 0.48, 95% CI: 0.25–0.90, P = 0.02, Figure 3B ) and 3-year DFS (OR: 0.39, 95% CI: 0.25–0.61, P < 0.0001, Figure 3C ).

In our study, we grouped RFA (Radiofrequency Ablation) and MWA (Microwave Ablation) together under the term TA (Thermal Ablation). In some of the included studies, the specific modality of thermal ablation was not explicitly specified. Therefore, we analyzed TA as a separate subgroup. Subgroup analyses based on ablation modality (MWA vs RFA vs. TA) revealed patients in the RFA+LR group gained significantly shorter 1-year, 2 -year, 3-year OS and 1-year, 3-year DFS than patients in LR group ( Figures 5A–F ).

Four of the included studies compared the complication between the TA+LR group and LR group. There was no significant difference in the incidence of postoperative complications between the TA+LR group and LR group (OR: 0.88, 95% CI: 0.73–1.05, P = 0.15, Figure 6 ).

Sensitivity analysis was carried out via the leave-one-out approach. This analysis indicated that the exclusion of any single study did not significantly affect the pooled results on the overall survival results. The results of the meta-analysis were therefore concluded to be stable and reliable ( Figure 7 ).

We assessed the funnel plot of the total complications for any publication bias. The funnel plot was symmetrical, indicating a lack of publication bias (p= 0.851) ( Figure 8 ).

This meta-analysis, encompassing 3,084 patients across six high-quality retrospective studies, provides critical insights into the evolving multimodal treatment paradigm for multifocal CRLM. The core finding is that combined thermal ablation and liver resection (TA+LR) is comparable to liver resection alone (LR) in terms of long-term survival rates, with no significant differences in overall survival (OS) at all time points (1–3 years), consistent with new evidence from propensity-matched analyses. Importantly, major complication rates were similar between the two groups (OR: 0.88, 95% CI: 0.73–1.05, P = 0.15), reinforcing TA+LR as a viable option for marginally resectable patients and challenging the traditional hierarchy that prioritizes resection as the sole curative approach. While combined resection and ablation is not superior to liver resection alone, it provides an effective treatment choice for patients who are not candidates for extensive liver resection, especially those with anatomically challenging lesions or limited future liver remnant (FLR). However, the decision to combine thermal ablation with surgical resection should be individualized based on patient-specific factors, including tumor burden, liver function, and overall functional status. Furthermore, liver-sparing hepatectomy allows for repeat resection in case of intrahepatic recurrence, thereby increasing the possibility of salvage surgery. For advanced CRLM patients with multiple metastases where R0 resection is deemed impossible, the combination of thermal ablation and liver resection has expanded the boundaries of CRLM resectability, preserving liver parenchyma while broadening the scope of curative surgery.

It is worth noting that, although there were no significant differences in overall survival (OS) at various time points (1–3 years), disease-free survival (DFS) was worse in the combined treatment group. Potential mechanisms may include the following: First, residual micrometastases in the ablation zone may accelerate intrahepatic recurrence without immediately affecting the patient’s overall survival, as salvage treatments for local recurrence are effective (19). Second, the technical limitations of TA in achieving histologically confirmed margins—especially for deep-seated perivascular lesions—may allow residual microscopic disease, a key factor since R1 resection margins independently predict DFS in CRLM. Most CRLM patients will develop recurrent disease after the first local treatment (20). However, there are discrepancies in the results regarding intrahepatic local recurrence, with patients treated with combined resection and ablation at a disadvantage. Yet, a recent retrospective study found a weak correlation between overall survival and recurrence-free survival, and the development of recurrent CRLM after liver resection does not necessarily rule out the possibility of cure. When feasible, repeat local treatment for recurrent disease has shown similar survival rates to those after the first liver resection.

Subgroup analysis revealed a steeper DFS decline in non-European TA+LR cohorts than in LR cohorts, possibly reflecting regional differences in the sequence of adjuvant therapy. For instance, Asian centers typically administer neoadjuvant chemotherapy before the ablation-resection sequence, while European guidelines prioritize upfront surgery for resectable disease (21). Subgroup analysis also showed that DFS was significantly lower in the radiofrequency ablation (RFA) subgroup compared to the microwave ablation (MWA) subgroup, likely due to the heat-sink effect of RFA on perivascular lesions and sublethal thermal injury-induced phenotypic transformation of tumor cells. Previous studies have shown that microwave ablation is less affected by tumor location and tissue type because of its lower sensitivity to the heat-sink effect. Moreover, microwave ablation can produce a more predictable, larger, and spherical ablation zone in a shorter time (22). Chong et al. (23)recently published their randomized McRFA trial, confirming that microwave ablation is as safe and effective as radiofrequency ablation for small hepatocellular carcinoma (HCC) lesions. Qiang Zhu and colleagues found significantly better DFS with microwave ablation compared to radiofrequency ablation in propensity score-matched patient groups with small perivascular HCC (24). A recent meta-analysis found a significantly lower local recurrence rate (ASR) in laparoscopic microwave ablation (OR: 2.16) compared to laparoscopic radiofrequency ablation, with no significant difference in major complication rates (OR: 0.21) (25).

From a clinical practice perspective, the subgroup analysis of this study provides three optimization pathways: First, multidisciplinary decision-making should prioritize ensuring FLR>30%, using three-dimensional CT reconstruction and indocyanine green clearance tests to select TA indications, providing an effective treatment choice for patients who are not candidates for extensive liver resection (26–28). Second, a standardized ablation technique protocol should be established, with MWA being the first choice for lesions >2cm or adjacent to large blood vessels (within 3mm of the main portal vein), and intraoperative contrast-enhanced ultrasound should be used to verify ablation margins (>5mm) to reduce the risk of local recurrence. Third, a dynamic monitoring plan should be implemented, recommending CT/MRI combined with CEA testing (sensitivity 92.4%) every three months for the first two years after TA+LR to identify intervenable intrahepatic recurrence early (29, 30). Notably, geographical differences show that the strategy of using neoadjuvant chemotherapy combined with staged ablation may weaken the DFS disadvantage (OR=0.97 vs. 0.52 in the European cohort). This suggests that the synergistic effect of treatment timing and regional diagnostic and treatment guidelines (such as the differences between ESMO and CSCO guidelines) needs to be considered in decision-making (31, 32).

This study has several limitations. Most included studies were retrospective, and long - term outcome studies on combined resection and ablation are rare. The retrospective design also led to insufficient control of unmeasured confounding factors such as RAS/BRAF mutation status, and the lack of histopathological data from the ablation zone made it difficult to objectively assess margin sufficiency. We initially noted the potential for bias and residual confounding. To elaborate, RAS/BRAF mutations can significantly influence CRLM prognosis and treatment response. Without data on these mutations, our results might be confounded, and our understanding of treatment effectiveness could be limited. Second, there was considerable heterogeneity in the modalities and protocols of TA. The variability in ablation parameters, such as power settings and treatment duration, was notable across the studies. For instance, microwave ablation (MWA) energy output ranged from ≥60W in some studies to lower levels in others, while radiofrequency ablation (RFA) protocols also differed in terms of applied power and treatment time. These differences in ablation parameters can significantly influence treatment efficacy and safety.

To address these limitations, we recommend future studies incorporate stratification based on molecular profiles such as RAS/BRAF mutations. This could provide more precise treatment effect estimates and help patient identify subgroups that may benefit most from specific therapies. Future research should also have larger sample sizes, prospective designs, and stricter control of confounding factors. RCTs and cohort studies are needed to confirm our findings and clarify the role of combined resection and ablation in CRLM treatment.