Integration of Radical Intent Treatment in Colorectal Liver Metastases

Simple Summary

Colorectal liver metastases (CRLM) represent a frequent and heterogeneous clinical scenario with potential for long-term survival or even cure in selected patients. Advances in surgical techniques, systemic therapies, and locoregional approaches have expanded the boundaries of radical intention treatment. However, optimal patient selection, proper integration of multimodal therapies, and precise prognostic stratification remain major challenges. In this review, we provide a comprehensive update on the management of CRLM with curative intent, focusing on current evidence regarding resectability, surgical and non-surgical strategies, and systemic treatment in different disease settings. Special attention is given to emerging concepts such as molecular profiling, immunotherapy, and the role of liver transplantation.

Abstract

Colorectal liver metastases (CRLM) management remains a complex conundrum in the context of potential curable disease. The combination of systemic therapy and surgery, with overall survival outcomes up to 58% at five years, has become the gold standard. Locoregional therapies have gained evidence in complementing surgery or even substituting it in selected cases. Adequate patient selection is paramount, but prognostic models have certain limitations that prevent their full implementation in clinical practice. A plethora of prognostic factors exists, with variable evidence supporting their definitive role. Thus, CRLM management decisions frequently vary depending on multidisciplinary team experience and hospital access to systemic and locoregional treatments. Definition of resectability has evolved in recent years due to technical developments in surgical and non-surgical approaches. Complexity is added when trying to fully understand the integration between local and systemic treatment. Whereas evidence in the context of resectable disease has been attempted in several phase III trials, definitive conclusions regarding the best approach to potentially resectable disease cannot be drawn. In addition, liver transplantation has gained evidence and is proposed in selected patients, raising a challenge regarding its integration and wider implementation. In this review, current standards in the management of CRLM regarding patient selection, resectability, surgical and non-surgical locoregional strategies, as well as the best systemic approach are covered.

1. Introduction

The management of colorectal liver metastases (CRLM) has undergone a substantial transformation over recent decades, driven by advances in surgical techniques, perioperative care, systemic and liver-directed therapies, and multidisciplinary collaboration. While up to one-quarter of patients with colorectal cancer present with liver metastases at diagnosis, an increasing proportion are considered eligible for curative-intent strategies, including hepatic resection, locoregional therapies, or a combination thereof.

Hepatic resection remains the cornerstone of curative treatment, with five-year overall survival rates (OS) exceeding 50% in experienced centers [1,2,3]. However, the definition of resectability has become dynamic and multidimensional, now encompassing anatomical, functional, and biological variables. Advances in liver-directed therapies have further extended resectability criteria, enabling curative approaches even in complex bilobar disease.

The emergence of non-surgical modalities, including stereotactic body radiation therapy (SBRT), hepatic arterial infusion chemotherapy (HAIC), and transarterial radioembolization (TARE), has expanded the therapeutic armamentarium, especially for patients with recurrent or unresectable disease [4]. Simultaneously, prognostic stratification has evolved beyond classical clinical scores [5,6,7,8], integrating tumor molecular profiling [9], immunologic signatures, and volumetric assessments, although their routine application in clinical practice remains limited.

Most recently, liver transplantation has re-emerged as an option in highly selected patients with unresectable disease, supported by randomized evidence from the TRANSMET trial [10]. This paradigm shift challenges traditional treatment algorithms and necessitates refined selection tools [11]. Concurrently, systemic therapy has become an essential component of curative-intent management, whether as perioperative treatment in resectable disease, as conversion therapy in patients with borderline resectability, or as a gateway to potential curative options such as surgery or liver transplantation in initially unresectable cases. Despite growing evidence supporting its use, optimal patient selection, regimen choice, and timing remain key areas for refinement.

In this review, we provide a comprehensive overview of the current evidence on the curative-intent management of CRLM, including imaging assessment, evolving definitions of resectability, surgical and non-surgical strategies, patient selection, systemic therapy in resectable and potentially resectable disease, and the emerging role of liver transplantation.

2. Preoperative Imaging Assessment

The main goals of CRLM evaluation prior to consideration of a treatment strategy with radical intent are the correct characterization of hepatic metastatic disease and the exclusion of extrahepatic metastatic disease.

Computed tomography (CT) scan achieves a sensitivity of 70–95% and a specificity of 95% in the detection of hepatic lesions, with a reduced cost and high availability [12,13,14]. Its main limitation is the detection of lesions smaller than 10 mm, where it has a false-negative rate of up to 10% [15], especially in the post-neoadjuvant reevaluation, where chemotherapy-associated liver injury (steatosis, steatohepatitis, or sinusoidal obstruction) could limit its detection ability [16,17,18]. An additional advantage of CT scan, in contraposition to magnetic resonance imaging (MRI), is the detection of extrahepatic disease, where it has a reduced sensitivity (58–64%), especially in smaller lesions or peritoneal locations, and high specificity (87–97%) [19].

Magnetic resonance imaging has emerged as an alternative to CT in liver assessment, with a sensitivity of 81% and specificity of 97% for the detection of liver metastases [14,15]. Despite its superior accuracy to CT in small lesions, peritoneal locations, and in the context of chemotherapy-associated liver injury, the cost and limited widespread availability have prevented its establishment as the first choice [20,21].

Although fluorodeoxyglucose positron emission tomography (FDG-PET-CT) has a non-negligible sensitivity (80%) and specificity (92%) for the detection of liver metastases [22], its high false-negative rate (87%) [23], especially after chemotherapy, makes it necessary to take its results with caution. One possible explanation would be the alteration of hepatic cellular metabolism after cytotoxic chemotherapy [18,24]. It is in the detection of extrahepatic involvement where it reaches its highest performance (sensitivity, 91%; specificity, 98%) [22]. This ability to pinpoint the distant extension of the disease can lead to a change in therapeutic management in up to 25% of cases [22].

Given there is no infallible imaging modality, intraoperative assessment remains necessary to fully stage the extent of disease. Intraoperative ultrasonography (IOUS) and contrast-enhanced IOUS (CE-IOUS) have emerged surpassing CT and MRI for the detection and control of known metastases. With real-time, high-resolution imaging, especially in small lesions (<1 cm), it should be routinely included in the surgical act as complementary techniques to cross-sectional imaging [25,26]. Regarding disappearing liver metastases (DLM), lesions initially diagnosed in pre-neoadjuvant staging and lost in follow-up because of treatment response, some series report a 67% detection rate intraoperatively through IOUS. Factors independently associated with its detection were moderate/severe hepatic steatosis and subglissonian location [27]. In addition, CE-IOUS could improve IOUS results in the detection of DLM (sensitivity 99%, positive predictive value (PPV) 98%) [28]. In the pre-surgical scenario, MRI remains more reliable than CT in the assessment of disappearing liver metastases, with a negative predictive value of 0.73 and 0.47, respectively [29].

Overall, the choice of liver assessment techniques depends on the characteristics and timing of the imaging study. While CT is used as the first imaging modality due to its availability and performance in liver parenchyma and extrahepatic disease, its deficiencies in small lesions or after systemic treatment should prompt MRI assessment. For staging completion, FDG-PET-CT may be a useful modality for further patient selection.

3. Definition of Resectability

Improvements in surgical technique of liver metastases have transformed resectability into a dynamic concept, offering a continuous expansion in the characteristics required to conduct radical-intent surgery. As a result, studies assessing CRLM surgery have used different definitions of resectability, making it difficult to compare criteria among the literature. Apart from technical and operability criteria, the biological behavior of disease, further detailed in the following sections, may influence the final indication for CRLM surgery.

Classically, liver disease has been classified into resectable, potentially resectable, and unresectable categories according to anatomical criteria. Resectable disease requires preservation of a 20% viable future liver remnant (FLR) with adequate biliary drainage and vascular inflow and outflow to obtain complete resection. However, FLR objectives vary depending on the quality of the liver parenchyma. Thus, while 20% FLR should be sufficient in a healthy liver, at least 30% or 40% may be necessary in chemotherapy-associated liver injury and cirrhosis, respectively [30]. When FLR is the only factor preventing a patient from curative resection, thus defining potentially resectable disease, additional techniques can be proposed to enhance healthy liver parenchyma (portal vein embolization, two-stage hepatectomy, associating liver partition and porta vein ligation for hepatectomy).

In summary, as a dynamic concept, liver resectability warrants proper assessment in a multidisciplinary team board, setting the possibility of resection before establishing a treatment plan. Given recent technical advances and the survival benefit of radical resection, every liver disease could be potentially resectable if biliary drainage and vascular flow in the FLR are assured.

4. Patient Selection

Broadening the definition of resectability and determining final surgery indications depending on multidisciplinary team expertise could increase the number of patients offered radical treatments, especially in experienced centers. However, the understanding of disease biology prompts adequate patient selection to enhance the results offered by surgery and avoid resection in poor-prognosis context.

4.1. Patient-Related Factors

Functional status and comorbidities may limit surgery eligibility, especially when perioperative systemic treatment is intended. Unfortunately, despite functional status and typical exclusion criteria common to any clinical trial protocol (e.g., heart disease within six months before surgery, a second primary tumor in the last five years prior to inclusion, or treatment with certain drugs depending on the nature of the study in question), thorough analysis of how patient-related factors may influence liver surgery outcomes remain limited. For example, tools such as the Child–Pugh classification or the Model for End-stage Liver Disease (MELD) score can provide guidance on the risk of mortality associated with liver resection. Depending on the anticipated extent of resection, patients with Child–Pugh B or a MELD score between 10 and 15 points could be operable, provided that an adequate pre-surgical optimization of comorbidities is ensured [31].

Given demographic aging and the higher incidence of colon cancer in older patients, adequate geriatric assessment using validated models should be used to enhance outcomes in this population, which faces an increased risk of surgical morbidity and antineoplastic treatment-related toxicity. It is observed that the feasibility of liver surgery may make clinicians more prone to indicate resection in older patients with low disease burden and prolonged disease-free intervals (DFI), preventing this fragile population from receiving potential toxic systemic therapy [32]. Nevertheless, most studies underlying current clinical practice guidelines, with some exceptions, restrict inclusion to patient under 75–80 years of age and with an ECOG score of no greater than 1 [33,34], conditions which are relatively frequent in routine clinical practice.

4.2. Cancer-Related Factors

Since the first publication of the prognostic scoring system proposed by Nordlinger et al. [7], several clinical prognostic models have attempted to predict outcomes after resection of CRLM (

Table 1. Prognostic models.

Score	Criteria	Risk
Nordlinger [7]	Age > 60 years Serous invasion of primary tumor (>pT3) Primary locoregional lymph node involvement Disease-free interval < 24 months Number of liver metastases > 3 Maximum diameter of liver metastases > 5 cm	Low: 0–2 Intermediate: 3–4 High: 5–6
Fong [5]	Primary locoregional lymph node involvement Disease-free interval < 12 months Number of liver metastases > 1 Maximum diameter of liver metastases > 5 cm Preoperative CEA > 200 ng/mL	Low: 0–2 High: 3–5
Nagashima [6]	Serous invasion of primary tumor (>pT3) Primary locoregional lymph node involvement Number of liver metastases ≥ 2 Maximum diameter of liver metastases > 5 cm Resectable extrahepatic metastases	Low: 0–1 Intermediate: 2–3 High: ≥4
Konopke [8]	Number of liver metastases ≥ 4 Preoperative CEA ≥ 200 ng/mL Synchronous liver metastases

), assessing criteria related to disease burden, temporal nature of the disease, patient age, and CEA levels. Notably, one of the main constraints to the generalization of all these models is that they were developed in patients undergoing surgery without prior systemic treatment, which could be a plausible scenario in resectable disease but uncommon in potentially or unresectable CRLM.

As presented in Table 1, classical prognostic models have included clinical and pathological variables with different outcomes, but there is a lack in the assessment of the molecular biology of the disease. A first attempt to evaluate the role of the disease molecular profile was proposed in the RAS Mutation Clinical Risk Score model [35,36], which replaced determinants such as DFI, number of metastatic lesions, and pre-surgical CEA level from the Fong model [5] with KRAS mutational status, thereby improving the prognostic model’s discrimination capacity in an external validation cohort.

Margonis et al. confirmed the model optimization when the molecular nature of disease was incorporated. Thus, assessment of KRAS mutational status together with the combined evaluation of the size and number of liver metastases (Tumor Burden Score, TBS), in addition to other clinical variables (

Table 2. GAME score [9].

Criteria	Score	Risk
Primary locoregional lymph node involvement	1	Low: 0–1 Intermediate: 2–3 High: ≥4
Preoperative CEA ≥ 20 ng/mL	1
Extrahepatic liver disease	2
KRAS mutation ¥	1
TBS2 * 3–8	1
TBS2 ≥ 9	2

* TBS² = (maximum metastases diameter) ² + (number of liver metastases) ². ^¥ Exon 2 (codon 12 y 13), exon 3 (codon 61).

), improved the discrimination capacity of the model with respect to the Fong score (Harrel’s C index 0.645; 95% CI 0.598–0.692 vs. 0.578; 95% CI 0.530–0.625; p = 0.008). The model, called GAME score, was validated in an external cohort of patients at Memorial Sloan Kettering Cancer Center, where it also demonstrated greater discrimination with respect to the model proposed by Fong et al. [9]. It should be noted that 66.7% of the patients analyzed had received cytotoxic treatment prior to liver surgery, in contrast to previously published clinical models. GAME score categorization into three risk groups (low, intermediate, and high) allowed statistically significant discrimination of OS in the patients included, with survival rates of 73.4%, 50.6%, and 11% at five years, respectively.

Despite improvement, some authors question the predictive capacity of these models. Bolhuis et al. [11] performed an external validation in 1105 patients from the Netherlands Cancer Registry with CRLM treated locally (resection and/or ablation) between 2015 and 2016. Disease-free and OS outcomes were analyzed, classifying patients according to the Fong and GAME models and predefining two subgroups regarding age (70 years old) and previous systemic treatment. The authors observed that, although the models discriminated several risk groups with statistically significant survival differences, the predictive ability of these models was insufficient (Harrel’s C index of 0.577; 95% CI 0.554–0.601, and 0.596; 95% CI 0.572–0.621 for the Fong and GAME models, respectively).

In the post-surgical setting, the HICAM study [37] retrospectively analyzed 176 patients who had undergone surgery for CRLM (up to 16.5% of patients had extrahepatic disease). Among the histological characteristics of the patients, it was observed that the so-called immunologically deserted phenotype, defined as the absence of lymphocytes in the analyzed tissue, had a statistically significant impact in terms of overall survival, contrary to the immune excluded and inflamed phenotypes. As a result of this analysis, the HICAM score was developed, including pre-surgical CEA (≥20 ng/mL), primary tumor resection, TNM stage at diagnosis, molecular status, histopathological growth pattern, and immune phenotype. The model allowed observation of differences in prognosis between high- and low-risk groups, in terms of relapse-free survival (RFS) (8.4 vs. 20.4 months, p < 0.001) and OS (30.4 vs. 105 months, p < 0.001). Additionally, a GAME score validation was performed, which allowed discriminating two risk groups with statistically significant differences in terms of RFS (20.5 vs. 8.8 months, p = 0.005, for the low- and high-risk groups, respectively) and OS (67 vs. 37.8 months, p = 0.005). Despite not being useful to select patients before surgery, HICAM score offers a complete assessment of disease determinants to refine prognosis which could be of interest when considering adjuvant treatment and follow-up.

An additional point of view was proposed by Zeeuw et al., analyzing the role of total tumor volume (TTV) [38,39] in the CAIRO5 cohort of patients [40], and observing that neither the TBS nor RECIST 1.1. assessment discriminated against statistically significant differences in terms of RFS and OS. However, the change in TTV (milliliters) before and after initiation of systemic therapy proved to be an independent predictor of survival [41].

A variable not studied in the prognostic models already mentioned, but of known relevance in metastatic disease, is the location of the primary tumor. Proximal tumors seem to lead to worse results in terms of RFS and OS compared to distal tumors (3-year RFS: 15% vs. 27%, respectively, p < 0.001; 3-year OS: 46% vs. 68%, respectively, p < 0.001), results that are independent of KRAS mutational status [42]. Regarding locoregional hepatic lymph node involvement, its presence has traditionally contraindicated hepatic surgery. In such cases, lower survival rates have been reported (18% at five years) [43] compared to those without lymph node involvement, making their presence relevant. Survival results worsen in a directly proportional manner as the lymph node involvement distances itself from the main hepatic pedicle: involvement at the level of area 1 or the hepatoduodenal ligament/retropancreatic area confers OS figures at five years of up to 30%, according to data reported by Pulitano et al. [43]. In contrast, involvement at the level of area 2 or common hepatic artery/celiac trunk confers 5-year survival rates between 0% and 14% [43,44,45]. Para-aortic lymph node involvement is consistently associated with zero 5-year survival figures [43,44].

Overall, initial attempts to model survival risk in CRLM resection have been outperformed by scores incorporating newer determinants of the extension of disease (TBS, TTV) and/or part of the molecular profile of the tumor (KRAS). Despite these efforts, a proper evaluation of relevant determinants in advanced disease is lacking, especially regarding clinical variables (primary tumor location, locoregional hepatic lymph node involvement) or more infrequent molecular subtypes (BRAF^V600E, HER2^amp, MMRd, etc.). In addition, patient-related factors, such as functional status and comorbidities, are underrepresented in the literature. Thus, it would be relevant to prospectively assess all these determinants in a multicentric cohort of CRLM treated with radical intent (Figure 1).

5. Surgery

The accumulated experience in surgical technique (resections oriented by segments, intraoperative assessment by ultrasound and vascular control through vascular occlusion techniques, anesthesia with low central venous pressure, or perfected parenchymal dissection devices) has made possible a change in tendency in what is currently considered technically resectable [46].

The opportunity offered by liver surgery to achieve prolonged disease control and even cure might have led to a more aggressive surgical approach compared with initial liver surgery indications. Thus, the limitations related to disease burden (traditionally, more than four metastases, maximum diameter greater than 5 cm, or planned resection margin less than 1 cm) [46] have been solved thanks to the use of neoadjuvant systemic treatment, two-stage resection, or the combination with local ablative techniques. Portal embolization (PVE) has made it possible to overcome the limitation of an insufficient FLR, while the possibility of resection of extrahepatic metastases, mainly pulmonary, has offered 5-year survival rates between 25 and 35% in worse-prognosis cases [47,48].

Couinaud’s anatomical classification permits the location of CRLM and planning of its resection depending on the vascular and biliary drainage. According to its distribution, a major hepatectomy would imply the resection of at least three segments. In minor hepatectomies, especially those involving metastase less than 5 cm without vascular involvement and located in peripheral segments, minimally invasive liver resection (MILR) arises as an alternative to classical open surgery [49,50]. MILR offers a reduced bleeding risk but limits the ability of the surgeon to manually explore the liver surface. Reported surgical series describe fewer complications, transfusions, and hospital stays with MILR, with comparable surgical times [51] and survival results [52].

Enhancing FLR in cases with reduced healthy parenchyma due to tumor or underlying disease is possible through portal vein embolization (PVE), portal vein ligation (PVL), Yttrium-90 transarterial radioembolization (TARE), or the associating liver partition and portal vein ligation for staged hepatectomy (ALPPS) procedure. PVE is the most widely used technique, offering hypertrophy by diverting blood flow from the embolized segments to FLR, but it requires a 4–6-week period to maximize outcomes, which could entail a risk dropout (20%) due to tumor progression or inadequate results [53]. ALPPS involves parenchymal transection and PVL in the first stage followed by liver resection within two weeks. It allows rapid hypertrophy but is accompanied by a high morbidity rate (Clavien-Dindo ≥IIIb) [53], which reserves this technique to salvage treatment after PVE failure. PVL and TARE are less used techniques. PVL is the same as PVE, conducted intraoperatively in the first stage of a two-stage hepatectomy, and is associated with similar morbidity to PVE [54]. TARE is a complementary modality to PVE, with a dual role of hypertrophy and downstaging by antitumoral effect. Despite being considered mainly a therapeutic tool, it has been recently established as a FLR-enhancing technique, due to its hypertrophy-generating capacity through embolization [53]. Limited availability and poor hypertrophy results limit its standardization [55]. Recently, the combined embolization of portal and ipsilateral hepatic veins, called liver venous depletion (LVD), could allow more extensive resections with adequate margins by inducing greater and more rapid hypertrophy than PVE, without a detriment in morbidity or mortality as seen in the DRAGON1 trial [56].

The selection of FLR-enhancing technique depends on the center’s experience and patient/disease characteristics. While PVE remains the most used technique, there exists controversy about whether ALPPS should replace it as the first enhancing maneuver. The advantages of ALPPS are more rapid and profound hypertrophy, which allows a higher number of patients to be classified as resectable compared with conventional techniques such as PVE (90% vs. 57–78%). In addition, such hypertrophy is produced in lesser time than PVE (7 to 14 days), which, theoretically, would imply a smaller dropout rate due to tumor progression. In contrast, the caveats of ALPPS that do not permit it to replace PVE are a higher rate of complications and post-procedure mortality. These limitations emphasize the need for exquisite patient selection when FLR enhancing techniques are being considered [57], especially when no clear long term survival advantage has been documented in the series reported [58,59].

An important aspect of surgery in synchronous CRLM is the timing of resection: classical or primary first, liver first, and simultaneous approach. A management proposal of synchronous disease is depicted in Figure 2. Regarding outcomes, systematics reviews and meta-analysis have shown no differences between the primary- or liver-first approach [60], with similar 5-year OS rates (OR 1.15; CI 0.93–1.42; p = 0.19; I² 18%). Similar conclusions have been extracted from real-world evidence. Andres et al. offered real-world data from LiverMetSurvey, showing comparable survival outcomes between both approaches [61]. It seems clear that surgical decisions should be made in a multidisciplinary committee, with the main determinants of the decision being the presence of symptoms and the prognostic disease burden. While primary-first surgery would allow the resection of symptomatic primary tumors, the liver-first approach [62] permits treating CRLM in cases where primary tumor treatment would imply multimodal treatment, such as rectal cancers, or when multiple and/or bilobar liver metastases are present. The simultaneous approach is intended to limit the surgical and anesthetic exposure time, but its feasibility would depend on the complexity of the surgeries involved. Despite some systematic reviews have observed a survival benefit in terms of 5-year OS compared with the liver-first approach (OR 0.47; 95% CI 0.25–0.90; p = 0.02; I² 0%), questions regarding the “synchronous resectable metastases” arise in the studies included, and therefore this conclusion must be considered with caution [60]. The main drawback of the simultaneous surgical sequence relays on the increased number of perioperative complications, especially in high TBS cases [63,64]. No benefit has been observed when simultaneous and primary-first approaches were considered [60].

Once the treatment sequence is chosen, proper liver resection should be proposed considering the individual patient’s situation. When treating resectable disease, unilobar or bilobar, with no need for liver hypertrophy, parenchymal-sparing surgery (PSS) permits resection while maximizing the amount of healthy liver parenchyma preserved. This non-anatomic resection has reported similar oncologic outcomes to anatomic surgery with lower morbidity [65] and offers salvage therapy when liver recurrence occurs (up to 33%) [66]. One-stage hepatectomy is typically proposed in multifocal unilobar disease and can be conducted in combination with PSS with or without liver hypertrophy techniques when necessary. Finally, two-stage liver resection sequences a first surgical procedure in which the predicted FLR is cleared of disease and liver hypertrophy is promoted by PVE/PVL, with a second operative time where the remaining liver parenchyma is cleared [67]. The use of the regenerative capacity of the healthy liver tissue offers an opportunity of radical treatment in cases of bilobar disease, as stated in the surgical series reported [68].

An important prognostic factor after surgery is the resection margin. The objective has evolved in recent years to an accepted margin >1 mm to be considered an R0 [69]. Although suboptimal, R1 resections still would confer a survival benefit given curative-intent treatment offered to patients [70,71]. Of note, there is a prognostic difference between R1 parenchymal (R1par) and R1 vascular (R1vasc) margin. It has been hypothesized that the interposition of a vessel in the affected margin would prevent tumor dissemination. As reported in retrospective series, R1par would confer worse survival outcomes compared to R1vasc, which would not differ from R0 resections [72].

Liver Transplant

The favorable experience and efficacy of liver transplantation in patients diagnosed with hepatocellular carcinoma has led to exploration of its potential role in metastatic liver disease due to colorectal cancer, primarily in the context of unresectable disease, although some series of liver transplantation in patients with resectable disease have been published [73,74].

The evidence available until 2024 consisted of surgical series with a limited number of patients (n = 5–21 patients). According to these series, liver transplantation in patients with unresectable liver disease could result in overall survival rates at one, three, and fiver years of 95–100%, 68–83%, and 60–83%, respectively [73,74].

To date, the latest update of the ESMO guidelines for colorectal cancer, published in 2023, does not consider liver transplantation outside a research context [4]. In 2021, the working group of the International Hepato-Pancreato-Biliary Association published recommendations for considering liver transplantation, limiting the disease scenario to unresectable disease with at least six months of response to systemic therapy, at least one year since the presence of liver metastases, and without diagnosis of BRAF^V600E mutation, microsatellite instability, or Lynch syndrome. Prior to transplantation, the primary tumor must have been resected by R0 resection with no signs of locoregional recurrence. In addition, certain histologies, such as undifferentiated adenocarcinoma or adenocarcinoma with signet ring cell component, should be excluded due to their poorer prognosis. Other factors, such as tumor metabolic volume (>70 cm³) or CEA level (>80 μg/L), were proposed as relative contraindications to the procedure [75].

It is important to note that the benefits suggested by the available series must be weighed against the risks and limitations inherent in organ transplantation. Although generally less than 10%, there are potential serious complications that can complicate the prognosis of patients (transplant rejection, bile leakage, hepatic or portal artery thrombosis). In addition, the immunosuppression required after the procedure can add to infectious, endocrinological, and renal complications, and even increase the risk of secondary neoplasms [76].

Recently, the TRANSMET study [10] provided the first randomized evidence supporting liver transplantation therapy in unresectable disease. In this study, Adam et al. randomized 94 patients between 2016 and 2021 to treatment with liver transplantation and chemotherapy vs. chemotherapy alone. The inclusion criteria established were, among others, age 65 years or younger, ECOG performance status 0–1, no detection of BRAF^V600E mutations, and R0 resection of the primary tumor. In addition, patients had to have had stable disease or partial response according to RECIST 1.1 criteria for at least three months and after no more than three lines of systemic treatment. One piece of data that emphasizes the need for adequate patient selection for this type of therapy is the fact that only 40% of the patients evaluated for inclusion were ultimately randomized. The profile of patients treated in the study included a population with proximal tumor disease (15%), KRAS mutation (20%), and, in all cases, metastatic disease was considered synchronous (0–1 months). In each treatment group, up to 70–80% of patients were randomized after one or a maximum of two lines of treatment. The primary endpoint of the study reached pre-specified statistical significance with a median overall survival not reached at 5 years of follow-up (57%) in the transplant arm versus 30 months in the control arm (13% at 5 years) (HR 0.37; 95% CI 0.21–0.65; p = 0.0003). After a median follow-up of 50 months, 72% of patients randomized to liver transplantation had a relapse, with up to 54% of these being pulmonary in location, and rescue therapy by surgery or ablation being possible in 46% of cases.

Although the TRANSMET study may provide sufficient evidence to add another therapeutic tool for the unresectable population, it also has certain limitations that may hinder its translation into daily clinical practice. First, organ shortages may limit access to this therapy depending on the patient’s context. Second, patient selection will be crucial to replicating the results obtained, and questions remain about the strategy to be followed for metachronous disease or other molecular profiles.

Surgical management of CRLM has evolved considerably, driven by advances in anatomical liver resection techniques, perioperative care, and liver-directed strategies aimed at expanding resectability criteria. Traditional constraints, such as high tumor burden or insufficient FLR, are increasingly overcome through systemic therapy, two-stage hepatectomy, or adjunctive hypertrophy techniques. Minimally invasive liver resection (MILR) has emerged as a viable alternative to open surgery in select patients, offering reduced morbidity with comparable oncologic outcomes. Parenchymal-sparing strategies and individualized sequencing, whether liver-first, primary-first, or simultaneous, enable safe resections even in complex bilobar disease. Importantly, the goal of R0 resection has shifted toward accepting margins >1 mm, with nuanced understanding of R1 subtypes (parenchymal vs. vascular). Lately, in highly selected patients with initially unresectable disease, liver transplantation has re-emerged as a viable option.

6. Non-Surgical Locoregional Treatments

The toolbox of non-surgical locoregional treatments in CRLM comprises local ablation techniques (TA, SBRT) and intra-arterial therapies (TARE/SIRT, HAIC, TACE) [4].

Regarding thermal ablation, the current recommendation is its use in unresectable disease or recurrent disease after a previous attempt at resection of liver metastases [4]. The two available TA modalities, radiofrequency or microwave, have shown similar results and safety profiles. However, microwave ablation may be the treatment of choice in cases that do not meet the criteria for radiofrequency ablation (fewer than three lesions, maximum diameter of each less than 3 cm, proximity to risky vascular structures) [77].

Radiofrequency ablation in combination to FOLFOX was tested in the CLOCC study, demonstrating an OS benefit in favor of the combination arm (43.1% vs. 30.3% at five years; HR 0.58, 95% CI 0.38–0.88), with a postoperative complication rate of 5.9% [78]. More recently, the non-inferiority phase III COLLISION trial compared surgery vs. TA in resectable CRLM. After a median follow-up of 28.8 months, the study was closed due to futility, observing no difference in terms of OS or PFS, with a morbidity and mortality profile favorable to thermal ablation, which poses TA as a reasonable alternative to surgery [79].

The ease of administration, its adequate integration into the systemic treatment strategy, the accumulated experience in oligometastatic disease, and the reduced profile of adverse effects are some of the reasons that have established SBRT as a therapeutic option used to treat technically unresectable liver metastases or in inoperable patients [80]. It is worth noting that the evidence for the use of SBRT comes from phase I and phase II studies and, occasionally, retrospective series. Local control achieved with SBRT is estimated to be 50–95% at one year, which could translate into OS figures of 16–32 months [80,81,82,83,84,85]. The heterogeneity of the populations treated in these studies limits the accurate interpretation of the results, as they include patients with different molecular and clinical disease profiles. It should also be remembered that the patients usually included in these studies for treatment with SBRT have been patients considered unresectable or inoperable from a surgical point of view, which may have weighed down the results in a population with a supposedly worse prognosis. Although the published evidence is not conclusive [86], the molecular profile could impact on the efficacy of SBRT: mutations in KRAS and TP53, especially when coexisting, appear to confer a poorer local control rate after treatment with SBRT [87].

Transarterial radioembolization is currently recommended as a therapeutic option when available systemic options have been exhausted [4]. The use of Y-90 microsphere TARE has demonstrated benefit in refractory and second-line disease in terms of ORR, PFS, and liver-PFS [4,88]. Nonetheless, the attempt to move the therapy to a possible first-line (RFA/FOLFOX) was explored unsuccessfully in three phase III clinical trials (SIRFLOX [89], FOXFIRE, and FOXFIRE Global [90]). Aggregated data from the three studies confirmed the absence of benefit in terms of OS compared to FOLFOX in metastatic disease confined to the liver [90].

HAIC exploits the preferential hepatic arterial supply of liver metastases and the liver’s first-pass effect to achieve high intrahepatic drug concentrations while minimizing systemic toxicity. Fluorodeoxyuridine (FUDR) is the most used agent, though other regimens (e.g., oxaliplatin, mitomycin, and more recently DEBIRI) have also been explored [91]. Hepatobiliary toxicity remains the main adverse effect. Clinical experience with HAI spans three therapeutic settings: adjuvant therapy post-metastasectomy, cytoreductive therapy, and palliative treatment. Although early studies failed to demonstrate survival benefits in the adjuvant context after metastasectomy—partly due to non-contemporary systemic therapy used [92]—more recent retrospective data from MSKCC (n = 2368) show a significant OS advantage for patients receiving adjuvant HAIC with FUDR in combination with modern chemotherapy (67 vs. 47 months; HR 0.67, p < 0.001) [93], sustained at 10-year follow-up (38% vs. 23.8%) [94]. However, due to the retrospective and single-center nature of the evidence, HAIC has not been adopted as standard therapy. In the cytoreductive therapy scenario, a phase II MSKCC study, combined HAIC with systemic chemotherapy yielding a 52% conversion rate to resectability, with median PFS of 13 months and OS of 38 months. Conversion was the only independent predictor of survival [95]. Considering palliative treatment, the CALGB 9481 trial demonstrated superior 2-year OS with HAIC-FUDR versus systemic 5-FU/LV (51% vs. 35%, p = 0.0034), despite worse control of extrahepatic disease [92]. This led to subsequent strategies combining HAIC with systemic chemotherapy to address micrometastatic progression.

Based on a statistically significant benefit in OS (22 months vs. 15 months, p = 0.031) communicated by Fiorentini et al. [96], DEBIRI TACE has been used in refractory liver disease. Its transition to first-line treatment was explored in 70 patients randomized to DEBIRI/FOLFOX vs. FOLFOX ± bevacizumab. In this phase III study, no statistically significant differences in PFS were observed between the treatment arms (15 months vs. 12 months, p = 0.18), with a higher rate of grade 3 or higher adverse events in the experimental arm (80% vs. 60%, p = 0.03) [97]. The latest update of the ESMO guidelines contextualizes the use of TACE as an option to consider in the non-curative setting [4].

As a whole, locoregional strategies offer radical-intent treatment without the surgical removal of CRLM. This is a relevant concept to have in mind when re-assessing tumoral response. Indeed, radiologically persisting metastases can express response characteristics that fall beyond the ability of size-based RECIST criteria normally used to measure treatment efficacy. Modified RECIST and EASL criteria assess residual tumor enhancement to predict viable tumor tissue. While widely used in hepatocellular carcinoma, its role remains to be elucidated in CRLM [98]. Choi and tumor attenuation criteria, which combine size and tumor density on CT scan, have demonstrated better correlation and PFS prediction when TARE is applied [99]. Finally, some of the literature proposes metabolic evaluation of the treated metastases through functional imaging criteria (EORTC PET, total lesion glycolysis (TLG) FDG-PET-CT) as a complementary assessment to morphological RECIST criteria, showing better sensitivity in locoregional treatments.

Overall, non-surgical locoregional therapies represent useful strategies in the therapeutic toolbox of CRLM, particularly in unresectable, recurrent, or chemotherapy-refractory settings. While integration with systemic treatment and advances in technique have expanded their clinical applicability, the current evidence base is largely derived from phase I/II trials or retrospective analyses, with limited data from randomized phase III studies. These limitations underscore the importance of appropriate patient selection and highlight the need for further prospective research to better define the role of these strategies in a multidisciplinary, personalized treatment approach. Recently, TA has been proposed as a useful procedure in the resectable scenario, with comparable OS rate and reduced morbidity.

7. Systemic Treatment

7.1. Perioperative Treatment in Resectable Disease

There is no consensus on the approach to patients with initially resectable CRLM. While some authors, based on the lack of proven benefit in OS, advocate for immediate surgery [100,101], others propose initiating systemic treatment in certain contexts of resectable disease with poor prognosis (≥4 metastases, elevated CEA, short ILE, bilobar disease, mutated RAS/BRAF^V600E) [4]. The main reason for the latter alternative, beyond verifying the efficacy of systemic treatment, would be to test the nature of the disease before surgery, attempting to identify rapidly progressive disease that would not benefit from immediate surgical intervention. If the latter alternative is chosen, it seems logical to limit the number of preoperative chemotherapy cycles and consider surgery as soon as its aggressiveness and response to treatment have been documented.

Despite the role of perioperative systemic treatment in resectable CRLM having been explored in several clinical trials, there is a constancy of limitations regarding those trials. It should be noted that, with the exception of the EPOC trial [102], studies in the context of resectable disease fail to analyze specific aspects such as size, number of metastases, the synchronous or metachronous nature of the disease, or its molecular characteristics to explore the potential benefit in specific subgroups of patients. Additionally, heterogeneity in the definition of resectability, different risk profile populations, protocol amendments, or imbalance in subsequent treatments have hampered the ability of those trials to reach solid conclusions. A common approach to systemic therapy has been proposed by EPOC investigators [102]. In this trial, perioperative treatment (three months preoperatively and three months postoperatively) did not show a statistically significant benefit in overall survival for the experimental arm with fluoropyrimidine and oxaliplatin doublet vs. surgery alone. Interestingly, a post hoc analysis of the EPOC trial showed a PFS benefit for perioperative FOLFOX4 in patients with elevated CEA (>30 ng/mL; HR 0.51, p = 0.03) and ECOG 0 (HR 0.61, p = 0.04) [102], possibly translating a more systemic character of the disease and the need for fitter patients to prevent discontinuation due to toxicity.

The role of antiEGFR was assessed in the NEW EPOC trial. The perioperative treatment with doublet chemotherapy and antiEGFR vs. chemotherapy alone did not show a statistically significant benefit in OS for the experimental arm. On the contrary, the trial showed a detrimental effect in OS for patients treated with antiEGFR therapy (81 months vs. 55.4 months; HR 1.45, 95% CI 1.02–2.05, p = 0.036), with an absence of benefit in imaging response or pathological resection status [34].

The use of bevacizumab in combination with chemotherapy in resectable liver disease has been questioned due to the risk of complications (impaired healing, thrombosis, bleeding). Loupakis et al. [103] analyzed the histopathological results in 42 patients previously included in phase II/III trials. Patients with available tissue samples who had received preoperative treatment with FOLFOXIRI or FOLFOXIRI-bevacizumab were selected and compared with 28 controls who had undergone immediate surgery. The pathological complete response rate was 16% vs. 11%, respectively, with no statistically significant differences (p = 0.685). Patients who received bevacizumab showed a higher tumor regression grade (TRG) (63% vs. 28%, p = 0.033), as well as a higher degree of necrosis (≥50%) (52% vs. 12.5%, p = 0.017). No differences were observed in liver toxicity after the use of bevacizumab.

In the adjuvant setting, several studies have attempted to evaluate the role of cytotoxic chemotherapy in resected liver disease [103,104], pointing to a possible benefit in terms of disease-free survival (DFS), but without demonstrating a benefit in terms of OS [105]. In this context, the most recent randomized evidence comes from the phase II–III JCOG0603 trial. In this study, 300 patients who underwent hepatectomy for metastatic colorectal liver disease were randomized to adjuvant FOLFOX6m or follow-up alone. After an interim analysis specified in the protocol, the study was closed based on benefit shown in DFS. However, at 5-year follow-up, a detriment in OS was observed for patients who had been treated with adjuvant therapy (83.1% vs. 71.2% at five years). Several reasons have been given to explain these results: lack of statistical power to detect OS differences, modification of the protocol during the study, chemotherapy-induced liver damage and its potential interference in the diagnosis of relapse, imbalance in subsequent treatments, greater toxicity and therefore lower adherence to the experimental arm, selection of clones with greater chemoresistance and consequent greater aggressiveness in relapse, and a population with a favorable risk profile (less than four metastatic lesions with a maximum diameter less than 5 cm) [106].

There are no phase III trials exploring the role of immune checkpoint inhibitors (ICI) in resectable CRLM, neither in the neoadjuvant nor perioperative scenario. As commented with further detail in the potentially resectable disease, the evidence in liver-only disease is extrapolated from the evidence assessing ICI in advanced disease, which suggests a limited benefit of antiPDL1 monotherapy in CRLM that could be overcome by dual PDL1-CTLA4 blockade [107,108].

Overall, in patients with resectable CRLM, upfront surgery remains appropriate for low-risk disease, while perioperative systemic therapy, based on FOLFOX, may be considered in cases with high-risk features. The association of directed therapy (antiVEGF, antiEGFR) is not supported by the published literature in this context of disease. Treatment decisions should be individualized based on disease biology, patient fitness, and multidisciplinary discussion.

7.2. Potentially Resectable Disease

The definition of potentially resectable disease refers to CRLM that could be resected with an R0 result but may imply technical and/or biological challenges that foster multimodal treatment. Despite its complexity, trials evaluating potentially resectable disease have used common criteria to define this clinical setting:

-

Impossibility of R0 resection, although some studies allow R1 resection [109].

-

Residual liver function after surgery less than 30% (less than 20% in healthy liver in some studies [110]), or with insufficient vascular supply and/or biliary drainage.

-

Contact with vascular structures that prevent resection (infiltration of all hepatic veins, both hepatic arteries, or both portal venous branches) [111,112].

-

Complex hepatectomy that does not allow the preservation of at least two contiguous hepatic segments.

The term conversion therapy has been coined to refer to the ability of systemic treatment to translate an initially unresectable disease into resectable, which is estimated in 7.3% (IQR 5–12.9%) depending on the disease profile and the treatment regimen used [113]. This conversion would translate into an increase in OS for initially unresectable disease, which would increase to 30–35% at five years, which is 10–15% higher than expected in this population if treated exclusively with systemic therapy [4]. Of note, trials investigating the role of conversion therapy have included patients who have been molecularly selected in a heterogeneous manner according to RAS status [114,115].

The choice of conversion systemic treatment should follow the same principles set out in the recommendations for the treatment of advanced disease [4], although the objective of tumor response becomes more relevant given the known correlation between the obtained imaging response and the surgical rescue rate (r = 0.96, p = 0.002) [116]. Careful selection of perioperative treatment and its duration, limited to the point of conversion, is recommended to reduce possible chemotherapy-induced liver damage that could prevent surgery [69].

In the KRAS exon 2 non-mutated population, the phase II CELIM and PLANET-TTD clinical trials compared the activity of fluoropyrimidine-based doublets with antiEGFR. Since all treatment arms received targeted therapy, the information provided by the studies relates to the activity of regimens with oxaliplatin or irinotecan. The overall response rate (ORR), the primary endpoint of both studies, was numerically superior in the oxaliplatin arms, without statistically significant differences. The resection rate was approximately 30% of patients treated in each arm of the two studies [111,114].

The question on whether antiangiogenic therapy adds efficacy to a fluoropyrimidine doublet was assessed in the BECOME [115] trial. With primary endpoint of conversion rate to surgery/complete resection, and a molecularly selected population, it demonstrated a statistically significant difference in directed therapy when associated with doublet chemotherapy (16–18% rate increase), which translated into a significant increase in OS.

Direct comparison between antiEGFR and antiVEGF directed therapy could be extracted from the ATOM trial [114], in which the addition to FOLFOX demonstrated an increased ORR favoring antiVEGF therapy (68.4% vs. 84.7%, for antiEGFR and antiVEGF, respectively, p = 0.0483). Despite this difference, no statistically significant differences were observed in PFS, the primary endpoint of the trial. It is important to note that only KRAS native patients were included. Based on more recent evidence, such as the phase III PARADIGM clinical trial, it is possible to ultra-select patients for antiEGFR treatment to optimize efficacy results in terms of response rate [117].

The role of triplet chemotherapy regimens has been explored in clinical trials with different designs, therapeutic schemas, molecular selection, and contexts of disease. The approach proposed by phase II OLIVIA [110] and METHEP [118] trials consisted of FOLFIRINOX regimen vs. doublet chemotherapy, combined with antiVEGF in OLIVIA protocol. Of note, no molecular selection was conducted in either of the trials, and METHEP protocol permitted the inclusion of up to three resectable metastatic lung lesions smaller than 2 cm. After follow-up completion, the primary endpoint of resection rate was 57–61% vs. 33–49% in favor of triplet combination treatment.

The PRODIGE14-ACCORD 21 study [119], with a phase II design, compared the efficacy of doublet vs. triplet chemotherapy combined with targeted therapy guided by the molecular profile related to KRAS status. The resection rate (R0 or R1), the primary endpoint, was 45.2% vs. 56.9% (p = 0.062), favorable to the triplet arm. No statistically significant differences were observed in terms of resection rate (R0 or R1) in patients treated with bevacizumab (mutated KRAS) or cetuximab (non-mutated KRAS) (44.7% vs. 55.6%, p = 0.087). With a median follow-up of 22.5 months, statistically significant differences were obtained in terms of OS (median of 36 months vs. not reached (p = 0.048), for the double and triplet arms, respectively).

Finally, the open-label phase III CAIRO5 randomized patients with right-sided colon cancer and/or RAS/BRAF^V600E mutation in initially unresectable CRLM to receive FOLFOXIRI-B vs. FOLFOX/FOLFIRI-B for a maximum of 12 cycles followed by 5-FU-B maintenance in patients in whom local hepatic therapy (surgery and/or ablation) could not be performed. With a median follow-up of 51.1 months, PFS was 9 months vs. 10.6 months in favor of the FOLFOXIRI-B arm (HR 0.76; 95% CI 0.6–0.98; p = 0.032). Right-sided and/or RAS/BRAF^V600E mutated median OS was 23.6 vs. 24.1 months for FOLFOX/FOLFIRI-B and FOLFOXIRI-B, respectively (HR 0.90; 95% CI 0.70–1.17; p = 0.44). For left-sided and RAS/BRAF^V600E wild-type tumors, median OS was 39.9 vs. 38.8 months with FOLFOX/FOLFIRI-B and FOLFOX/FOLFIRI-P, respectively (HR 0.95; 95% CI 0.68–1.32; p = 0.75). The ORR was higher in the FOLFOXIRI-B arm (54% vs. 33%; p = 0.0004). As a result, the complete local treatment rate was higher for the FOLFOXIRI-B arm (51% vs. 37%; p = 0.013), at the cost of increased grade 3 or higher toxicity, mainly due to neutropenia (39% vs. 13%) and diarrhea (20% vs. 3%), and a higher rate of postoperative complications (27% vs. 15%) [40]. This highlights the need for adequate patient selection if treatment with triple therapy associated with bevacizumab is considered.

Patients with tumors harboring the BRAF^V600E mutation exhibit significantly reduced OS compared to their wild-type counterparts, which has been a constraint for CRLM surgery in this subgroup [120,121]. The recently reported results from the BREAKWATER study demonstrated a survival benefit, both in OS (30.3 vs. 15.1 months; HR 0.49, 95% CI 0.38–0.63) and PFS (12.8 vs. 7.1 months; HR 0.53, 95% CI 0.41–0.68), when encorafenib and cetuximab were added to mFOLFOX6 regimen. In addition, patients treated with the experimental arm showed an improved ORR (60.9% vs. 40%) [122,123]. The results of this trial in the CRLM subgroup have made clinicians reconsider the indication of radical intent treatment in BRAF^V600E tumors.

The published literature does not resolve the controversy regarding the role of ICI in potentially resectable disease. Despite limited studies having shown a high complete pathological response rate (92%), even in the presence of persistent imaging evidence of disease (86%), some studies point to a lower benefit of immunotherapy in metastatic disease with liver involvement [122,123], especially when ICI monotherapy is chosen, as was the case in the KN-177 trial [107]. Several mechanisms related to the hepatic microenvironment have been postulated to explain the different effects of ICI on liver metastases compared to extrahepatic dissemination. First, the immunotolerance inherent to the liver might suppress the antitumoral response via CD8+ T-cell apoptosis through macrophage activation [123]. Additionally, Treg cells and their IL10 secretion would reduce PDL1 expression in monocytes, thus inhibiting CD8+ activity and resulting in an immune deserted’ phenotype [124,125].

Conversely, the CheckMate 8HW showed that the addition of CTLA4 blockade offered a maintained PFS benefit even in the liver metastases population [108]. In any case, the benefit seems to be restricted to dMMR/MSI-h tumors, with an absence of it in proficient MMR adenocarcinomas.

Overall, maximizing the potential effect of preoperative systemic treatment with the aim of conversion is a reasonable strategy that offers a resection rate of around 50–60%, at the cost of significant toxicity. Despite the lack of universal evidence-guided recommendations in the potentially resectable disease, some assumptions might be proposed. The addition of antiEGFR therapy in the case of non-mutated RAS and BRAF^V600E status offers superior conversion results. The use of antiVEGF may increase response rates when combined with chemotherapy, which may translate an OS benefit. Current guidelines recommend the use of chemotherapy in combination with antiEGFR in non-mutated RAS and BRAF^V600E tumors located on the left side when the goal is complete resection. In cases of mutated RAS or tumors located on the right side, bevacizumab therapy associated with FOLFOXIRI or a chemotherapy doublet would be the treatment of choice, depending on the patient’s tolerance to the toxicity of the triplet, with the use of antiEGFR therapy being acceptable in right-sided disease if the goal is cytoreduction [4]. Finally, as a consequence of the BREAKWATER trial, the improved survival results in BRAF^V600E disease observed with doublet chemotherapy combined with cetuximab and encorafenib should not prevent this population from surgery solely on the basis of a poorer prognosis [126].

7.3. Unresectable Disease

Approximately 80% of patients with CRLM will be unresectable due to the number, size, or location of the metastases, the presence of extrahepatic disease, or patient conditions [127]. In these cases, the treatment of choice will be systemic treatment guided by the molecular profile of the disease, the location of the primary tumor, and the patient’s comorbidities [4]. Keeping in line with current guidelines recommendations, single- or double-agent chemotherapy plus antiVEGF would be the treatment of choice in MMRp right-sided and/or RAS wild-type disease. When left-sided RAS and BRAF wild-type tumors are present, treatment regimen should include chemotherapy and antiEGFR therapy. The presence of BRAF^V600E mutation has been recently addressed in the BREAKWATER trial, positioning mFOLFOX6 in combination with encorafenib and cetuximab as first choice in this setting. In MMRd cases, pembrolizumab or nivolumab–ipilimumab are recommended in the first line [4].

Consistently, in palliative chemotherapy clinical trials, there are 10–15% of patients considered to have unresectable disease, despite this, end up undergoing liver surgery. According to more contemporary data, the proportion of CRLM that could be ‘converted’ to resectable disease would reach 24–39% of cases, varying by population, regimen used, and the addition of locoregional treatments [128]. Potential conversion rate predictors have been proposed to better define treatment goals in therapy. Reduced size and number of metastases, left-sided primary tumors, absence of nodal involvement, RAS/BRAFV600E wild-type, low pre-treatment CEA and CA 19.9 levels, and radiographic response to systemic treatment correlate with the probability of reaching conversion [129].

These are often patients who, due to the presence of metastatic disease, a high disease burden, or the location of liver metastases, were considered unresectable and who, after one or more systemic treatments, with or without local treatments for liver metastases or metastases in other locations, end up opening the opportunity for resection of liver metastases that had not initially been considered. Regarding unresectable disease limited to the liver, the most relevant information in unresectable disease comes from the TRANSMET trial. As commented previously, its results endorse the feasibility of liver transplantation in highly selected patients [10].

8. Conclusions and Future Directions

The management of colorectal liver metastases (CRLM) with curative intent has evolved into a highly complex, multidisciplinary endeavor. Surgical resection remains the cornerstone for long-term survival in resectable disease, supported by advancements in parenchymal-sparing techniques, liver hypertrophy strategies, and minimally invasive approaches. However, patient selection based on prognosis is paramount, while risk score discrimination remains affected by heterogenic variables and, often, non-contemporaneous treatments regimens. Thus, it would be desirable to reassess the published evidence incorporating relevant clinical, molecular, and therapeutic variables not yet considered in the CRLM scenario.

The role of perioperative chemotherapy in resectable disease is nuanced and should be reserved for selected high-risk patients. In patients with potentially resectable disease, systemic therapy guided by molecular and anatomical characteristics can enable surgery in a substantial proportion of cases, though toxicity and optimal sequencing remain challenges. Even in cases of unresectable disease, the opportunity for potential curative-intent treatment for liver metastases should be considered after systemic therapy and with the emergent role of liver transplant. Further clinical trials containing novel therapies (ICI, therapies directed at BRAF^V600E mutations or antiHER2 treatments) should offer insight into whether the proposed algorithm to CRLM treatment remains valid in these specific populations or not.

Non-surgical locoregional therapies, such as thermal ablation and hepatic arterial infusion, have proven valuable, particularly in non-operable or recurrent settings, but require further validation through prospective trials. Moving forward, precision oncology—through the integration of molecular profiling, imaging biomarkers, and refined prognostic tools—will be key to personalizing treatment and expanding access to curative strategies across the spectrum of CRLM.