Immunotherapy in liver metastases: Challenges, emerging evidence, and future directions

Abstract

Liver metastases are common in advanced cancer and are consistently associated with poorer survival and limited treatment options. Although immunotherapy has transformed outcomes for many malignancies, patients with liver involvement respond less well, highlighting the liver as a uniquely challenging site for effective antitumor immunity. In this review, we examine the biological features that make the liver especially permissive to metastatic growth and immune escape. We first describe the liver’s distinctive anatomy and immune environment, including its dual blood supply, specialized sinusoidal vasculature, and intrinsic immune tolerance. We then discuss how primary tumors actively condition the liver before metastatic seeding through pre-metastatic niche formation driven by tumor-derived cytokines, chemokines, extracellular vesicles, and metabolic signals. A major focus is the dynamic crosstalk between metastatic tumor cells and liver-resident populations, including Kupffer cells, hepatic stellate cells, hepatocytes, and liver sinusoidal endothelial cells. These interactions promote immune suppression, fibrosis, angiogenesis, and metabolic adaptation, while infiltrating immune cells such as myeloid cells, T cells, and natural killer cells are reprogrammed toward dysfunctional states that limit immunotherapy efficacy. We integrate these mechanistic insights with clinical data from cancers that frequently metastasize to the liver, highlighting consistent patterns of immunotherapy resistance and emerging combination strategies aimed at overcoming liver-specific barriers.

Key Words: Liver metastasis; Tumor microenvironment; Immune crosstalk; Cancer immunotherapy; Translational oncology

Core Tip: This review highlights how the liver’s unique vascular and immune environment fosters metastasis. We summarize key cellular interactions, the concept of the hepatic pre-metastatic niche, and emerging therapeutic strategies that may improve outcomes for patients with liver metastases.

INTRODUCTION

Liver metastases are a common and clinically significant challenge in patients with advanced malignancies[1]. Tumors originating from the gastrointestinal (GI) tract, such as those of the pancreas, stomach, and colon, as well as from extra-GI sites including the breast, lung, and skin, frequently metastasize to the liver due to its unique dual blood supply and immuno-tolerant microenvironment[2]. Despite therapeutic advances, liver metastases often confer a poor prognosis and remain challenging to treat effectively[3]. Management of liver metastases often requires a multimodal approach tailored to tumor origin and disease extent: Therapies include surgical resection, thermal ablation (radiofrequency or microwave), regional interventions such as transarterial chemoembolization, radioembolization, or hepatic arterial infusion, and systemic therapies including chemotherapy, molecularly targeted agents, and immunotherapy[4]. Immune checkpoint inhibitors (ICIs) have transformed cancer therapy, offering durable responses in several malignancies[5]. However, emerging data consistently show that patients with liver metastases derive less benefit from immunotherapy compared to those without hepatic involvement[3]. This discrepancy is observed across multiple tumor types and is supported by clinical trial subgroup analyses and meta-analyses[3]. The liver’s immunosuppressive microenvironment-characterized by cytotoxic T-cell and natural killer (NK) cell exhaustion and exclusion, along with the expansion of multiple immunosuppressive cell populations-impedes effective anti-tumor immunity[6]. Consequently, liver metastases may represent a distinct immuno-biological barrier to successful multimodality treatments[3]. Understanding these liver-specific mechanisms is critical for optimizing therapeutic strategies and improving patient outcomes. This review aims to synthesize current clinical evidence and mechanistic insights into immunotherapy resistance in liver metastases and to highlight emerging therapeutic approaches that may overcome this challenge. Because many hepatic immune mechanisms operate similarly across tumor types, these concepts are presented in unified mechanistic sections, with tumor-specific clinical patterns detailed separately later in the manuscript.

UNIQUE ANATOMICAL FEATURES OF THE LIVER AND THEIR IMPACT ON METASTASIS

The liver’s anatomy and microarchitecture create a distinct environment for tumor cell settlement. One of the most salient anatomical features is its dual blood supply: Approximately 75% of hepatic inflow comes via the portal vein (draining the GI tract), and 25% via the hepatic artery[7,8]. This configuration exposes the liver to a continuous influx of dietary antigens, microbial products, and-importantly-circulating tumor cells (CTCs) originating from both GI and non-GI primaries[9]. Because the portal venous network channels blood from abdominal organs first through the liver, it makes the liver a primary “first-pass filter” for tumor cells arising in colorectal, gastric, pancreatic cancers, and even extra-GI malignancies such as breast, lung, and melanoma[10].

However, intact portal flow appears to be a prerequisite for efficient metastatic seeding in hepatic sinusoids. In a large study of 1453 cirrhotic livers with liver masses, only 1.7% turned out to be metastases (24 cases), in contrast to the preponderance of primary hepatic tumors[11]. Among those metastases, the majority (71%) occurred in livers with mild to moderate fibrosis (Laennec stages 4A/4B) and patent portal veins; fewer cases occurred in severe cirrhosis (4C) with reversal or obstruction of portal flow[11]. These findings imply that disturbed portal hemodynamics and fibrotic remodeling impede CTC access to the hepatic parenchyma, acting as a hemodynamic and structural barrier.

This concept is reinforced by older post-mortem data showing lower rates of metastases in cirrhotic compared to non-cirrhotic livers: The odds ratio for metastatic incidence in cirrhosis was 0.47[12]. In other words, cirrhotic architecture-despite being pathologic-may confer a degree of “metastatic resistance” via altered flow and vascular remodeling.

Beyond macro-hemodynamics, the sinusoidal microarchitecture provides additional layers of complexity. Liver sinusoidal endothelial cells (LSECs) form the fenestrated lining of hepatic sinusoids, maintaining immune tolerance and vascular homeostasis through antioxidant and anti-inflammatory activity[13-15]. When injured, LSECs lose fenestrae and develop intracellular gaps, up-regulating adhesion molecules such as ICAM-1 and MMP9 that facilitate tumor cell adhesion and transmigration. The diverse functions of LSECs in liver metastasis will be further elaborated below in a separate subsection below[14,16,17].

The liver is uniquely organized into lobular zones with distinct oxygen, nutrient, and metabolic gradients that shape its physiological and pathological responses. Peri-portal hepatocytes, located near the portal triad, exhibit high oxidative metabolism and oxygen consumption, whereas peri-venous hepatocytes, situated near the central vein, reside in relatively hypoxic regions and rely predominantly on glycolysis for ATP generation[18]. This zonal heterogeneity establishes spatially defined micro-environments that can restrict immune infiltration and drug penetration, thereby contributing to the poor therapeutic response and high resilience of liver metastases compared with metastases at other sites[6,19]. To successfully colonize the liver, tumor cells must adapt to its hypoxic milieu through metabolic reprogramming. Under oxygen-deprived conditions, metastatic cells secrete creatine kinase, brain-type into the extracellular space, where it catalyzes the conversion of creatine to phosphocreatine[20]. The resulting phosphocreatine is then imported via the SLC6A8 transporter to regenerate intracellular ATP, supporting tumor survival and proliferation within the hypoxic liver niche[20]. Additionally, hyper-activation of hypoxia-inducible factor 1 promotes glycolytic adaptation through up-regulation of its downstream target pyruvate dehydrogenase kinase 1, further enhancing metastatic fitness in the liver microenvironment[21].

Together, these anatomical and micro-architectural features, including the liver’s dual blood supply, uniquely fenestrated sinusoidal endothelial structures, and lobular zonation with distinctive oxygen, nutrient, and metabolic gradients, create a relatively hypoxic yet highly specialized immuno-biological and hemodynamic environment that governs the formation and progression of liver metastases.

THE CASCADE OF LIVER METASTASES

Pre-metastatic niche

Pre-metastatic niche (PMN) refers to a permissive microenvironment established in distant organs prior to the arrival of CTCs, creating fertile “soil” for metastatic seeding and immune evasion[22-24]. In the liver, PMN formation is orchestrated by tumor-derived soluble factors, extracellular vesicles (EVs), and myeloid cell recruitment, resulting in distinctive molecular, stromal, and immunologic remodeling that distinguishes hepatic niches from those in the lung or bone.

Tumor-derived exosomes enriched with integrins-particularly αvβ5-selectively home to Kupffer cells (KCs), where they trigger transforming growth factor-beta (TGF-β) release, activation of hepatic stellate cells (HSCs), and fibronectin deposition. This extracellular matrix (ECM) remodeling recruits bone marrow-derived myeloid progenitors, establishing a fibrotic and immunosuppressive scaffold that supports later colonization[22,25]. In parallel, cytokines and chemokines such as CCL2, CXCL1, and CX3CL1 drive the accumulation of myeloid-derived suppressor cells (MDSCs), which dampen cytotoxic T-cell activity and promote regulatory T-cell expansion[26]. Within this evolving microenvironment, KCs undergo phenotypic reprogramming toward an M2-like, programmed death ligand-1 (PD-L1)⁺ state, driven by metabolic stress and tumor-derived signals, amplifying T-cell exhaustion and immune escape[27]. The liver’s intrinsic tolerogenic landscape-shaped by its dual blood supply and constant exposure to gut-derived antigens-further enhances these suppressive mechanisms, setting it apart from other metastatic sites[27,28]. Finally, liver-specific metabolic and fibrotic changes, including lipid accumulation, hypoxia, and collagen deposition, reinforce immune exclusion and create a “cold”, immune-resistant microenvironment that undermines the efficacy of immune checkpoint blockade in liver metastases[29,30].

Liver metastatic cascade and immune resistance mechanisms

The metastatic cascade is a multi-step biological process by which cancer cells detach from a primary tumor, survive in circulation, and colonize distant organs[31]. Although this cascade occurs across multiple metastatic sites, liver metastases exhibit unique molecular, immunologic, and microenvironmental traits that distinguish them from other organs[2,32].

Mechanisms of tumor invasion, intravasation, extravasation, and hepatic colonization

Tumor invasion and intravasation initiate when malignant cells degrade the ECM and breach vascular or lymphatic barriers. This process is mediated by matrix metalloproteinases (MMPs), epithelial-mesenchymal transition (EMT), and dynamic interactions with endothelial and stromal components that facilitate motility and vessel penetration[33]. Once within the circulation, CTCs encounter substantial hemodynamic shear stress and immune surveillance[34]. To evade destruction, CTCs frequently aggregate with platelets, forming protective micro-thrombi that shield them from neutrophil and NK-cell-mediated cytotoxicity and phagocytosis[35].

Within the liver, hemodynamic and structural features favor metastatic arrest. The hepatic sinusoids, characterized by low shear stress, fenestrated endothelium, and abundant adhesion molecules such as VCAM-1, ICAM-1, and selectins, provide an ideal microvascular niche for tumor cell adhesion[36-38]. Tumor extravasation is further enhanced by inflammation-induced up-regulation of adhesion molecules-particularly ICAM-1-on LSECs. Experimental and translational models demonstrate that ICAM-1 expression on LSECs induces tumor cells to secrete prostaglandin E2, interleukin (IL)-6, vascular endothelial growth factor (VEGF), and MMPs, which collectively promote endothelial transmigration and remodeling of the peri-sinusoidal matrix, ultimately facilitating hepatic colonization[37]. Following extravasation, the hepatic microenvironment initiates a unique sequence of immunosuppressive responses that differ from other organs (Figure 1 and Table 1).

Figure 1 Tumor microenvironment of liver metastases. Tumor cells interact with hepatocytes, Kupffer cells, hepatic stellate cells, liver sinusoidal endothelial cells, and infiltrating immune cells through cytokines, chemokines, and exosomes. These signals induce fibrotic remodeling, immunosuppression, and endothelial barrier formation. Kupffer cells, tumor associated macrophages, and myeloid-derived suppressor cells secrete interleukin-10 and transforming growth factor-beta, while hepatocytes and fibroblasts contribute growth and survival factors such as fibrinogen-like protein 1, insulin-like growth factor 2, and serum amyloid A1/A2. Collectively, these interactions create a fibrotic, immune-excluded niche that promotes tumor colonization and resistance to therapy. LSEC: Liver sinusoidal endothelial cell; TGF-β: Transforming growth factor-beta; IL: Interleukin; MDSC: Myeloid-derived suppressor cell; NK: Natural killer; NET: Neutrophil extracellular trap; CAF: Cancer-associated fibroblast; HSC: Hepatic stellate cell; IGF2: Insulin-like growth factor 2; HGF: Hepatocyte growth factor; FGL1: Fibrinogen-like protein 1.

Table 1 Liver-resident cells and their roles in metastatic colonization and immune suppression.

Cell type	Key functions in liver metastasis	Roles in metastatic colonization & immune suppression
KCs[39-46]	Liver-resident macrophages that are the first immune cells to contact CTCs in hepatic sinusoids	Can exert early tumoricidal activity but may become reprogrammed into an immunosuppressive phenotype promoting T-cell exhaustion and MDSC accumulation
LSECs[13-16,47-49]	Fenestrated endothelial cells regulating leukocyte trafficking, antigen presentation, and vascular tone	Under pathological activation, LSECs upregulate adhesion molecules and promote tumor cell arrest, transmigration, and angiogenesis
HSCs[50-55,109]	Perisinusoidal cells that can transition into myofibroblast-like CAFs and remodel ECM	Activated by tumor-derived factors to secrete IL-8, deposit collagen, and create a fibrotic, immune-excluded microenvironment
Hepatocytes[56-61]	Primary parenchymal liver cells central to metabolism and paracrine signaling	Modulate Fas/FasL, IGF2, HGF, and FGL1 to support tumor survival, proliferation, and suppression of CD8+ T-cell and NK-cell responses

KC: Kupffer cell; CTC: Circulating tumor cell; MDSC: Myeloid-derived suppressor cell; LSEC: Liver sinusoidal endothelial cell; HSC: Hepatic stellate cell; CAF: Cancer-associated fibroblast; ECM: Extracellular matrix; IL: Interleukin; FasL: Fas ligand; IGF2: Insulin-like growth factor 2; HGF: Hepatocyte growth factor; FGL1: Fibrinogen-like protein 1; NK: Natural killer.

Cellular orchestration of metastatic colonization in the liver

The liver’s complex architecture and diverse cellular ecosystem create a unique environment for metastatic colonization. Beyond hepatocytes, a variety of non-parenchymal cells-including KCs, LSECs, HSCs (Ito cells), and infiltrating immune cells-coordinate local immune, metabolic, and stromal responses that dictate whether disseminated tumor cells are cleared or colonize successfully. Each cell type contributes distinct yet interconnected roles: KCs and LSECs regulate immune surveillance and vascular adhesion, while stellate cells mediate fibrotic remodeling and extracellular-matrix deposition. Understanding how these hepatic cell populations interact and reprogram under tumor-derived signals is essential for identifying therapeutic opportunities to disrupt the formation of a pro-metastatic niche (Figure 1 and Table 1).

KCs

KCs, the liver’s resident macrophages, play a dual role in metastatic progression, initially anti-tumorigenic but later becoming immunosuppressive. Their activation is governed by a balance of activating and inhibitory receptors that normally protect hepatic tissue from excessive inflammation but can shift toward immune tolerance during metastasis. In early liver metastasis, KCs are highly phagocytic and eliminate tumor cells. However, this capacity declines at later stages due to induction of musculoaponeurotic fibrosarcoma oncogene homolog B and c-musculoaponeurotic fibrosarcoma oncogene homolog, which limit their tumor infiltration and phagocytosis[39]. The lineage-defining factor inhibitor of DNA binding 3 (ID3) maintains KC phagocytic competence and promotes recruitment and activation of NK and CD8⁺ T cells to suppress tumor growth; loss of ID3 skews KCs toward an immunosuppressive phenotype[40].

Once reprogrammed, KCs actively support metastasis. Tumor-derived deleted in malignant brain tumors 1 binds to mucin 1 on KCs, triggering nuclear factor kappa B activation and induction of chemokine (C-C motif) ligand 8 and L-selectin, which recruit neutrophils and promote neutrophil extracellular trap (NET) formation, establishing a pro-metastatic niche[41]. KCs can also undergo CD206⁺ M2-like polarization in response to tumor-secreted cytokines such as C-X-C motif chemokine ligand 16, secrete immunosuppressive mediators (IL-10, TGF-β), and express Fas ligand (FasL), inducing apoptosis of Fas⁺ CD8⁺ T cells-a mechanism distinctive to the hepatic metastatic environment (Figure 1 and Table 1)[42-45]. Given their inherent phagocytic capacity, KCs have emerged as a target for therapeutic re-activation. β-glucans, pathogen-associated molecular patterns that signal through Dectin-1, can reprogram macrophages toward anti-tumor activity. In murine models, treatment with soluble β-1,3/1,6-glucan (odetiglucan) combined with immune checkpoint blockade [anti-programmed cell death 1 (PD-1)] significantly prolonged survival after metastatic challenge[46].

LSECs

LSECs form the highly specialized lining of hepatic sinusoids, serving as the anatomical interface between the liver’s dual blood supply-the portal vein and hepatic artery-and the parenchyma[13]. Unlike continuous endothelial cells found in other organs, LSECs possess fenestrations that allow for efficient molecular exchange and maintain a non-inflammatory, tolerogenic environment[15,47]. Fenestrated LSECs exert antioxidant and anti-inflammatory functions, scavenging reactive oxygen species and producing nitric oxide, which helps preserve hepatic microvascular homeostasis[15,47]. They also suppress HSC activation, thereby preventing fibrogenesis and indirectly impeding metastatic colonization[14,48]. In contrast, non-fenestrated LSECs exhibit a marked increase in the expression of CD31, CD34, and molecules such as von Willebrand factor, which is accompanied with development of basement membranes and ultimate transformation into a continuous endothelium[16]. However, the role of LSECs phenotype switching, i.e., from fenestrated to capillarized LSECs, remains unclear in liver metastases. The evidence we discussed above showed the cirrhosis is a protective factor against liver metastases, which seems to suggested the capillarized LSECs prevents metastatic seeding. However, recent studies revealed that pathological liver injury induces intracellular gap formation in LSECs through the destruction of fenestrae, creating permissive entry points for tumor cells. Cancer cells stimulate IL-23-dependent TNF-α secretion from LSECs, leading to F-actin depolymerization and up-regulation of MMP9, ICAM-1, and CXCLs-molecular changes that facilitate tumor cell transmigration and metastasis (Figure 1)[49].

HSCs

HSCs, or Ito cells, which differentiate into myofibroblasts/cancer-associated fibroblasts (CAFs) and deposit dense collagen-rich ECM, are important cell types to suppress T-cell driven anti-tumor immunity and create “cold” tumor microenvironment (TME) by forming dense collagen-based stroma to hinder immune cell infiltration and to shield tumor cells from damages inflicted by many anti-cancer therapies[50]. The process cell-derived fibrosis is driven by complex molecular and immune mechanisms, and dynamic communications of tumor cells, HSCs and other cell types in the TME. In the liver metastasis of gastro-intestinal malignancies, the HSC-CAF conversion often starts off with tumor cell-derived signalings. Tumor cells can produce exosomes, or secrete soluble factors such as cytokines or chemokines, to influence HSC behavior. For example, AMIGO2-containing sEVs derived from gastric cancer cells actively modify the hepatic microenvironment by activating HSCs and inducing IL-8 secretion, which promotes gastric cancer cell migration into the liver parenchyma[51]. Exosomal sphingosine kinase 1 (SPHK1) increased the migration of colorectal cells, and activated HSCs by regulating pAKT[52]. For example, POU6F2 over-expressing gastric tumor cells promotes the conversion of HSCs into CAFs via transcriptional upregulation of insulin-like growth factor 2 (IGF2) and subsequent activation of PI3K/Akt signaling as well as[53]. Similarly, suprabasin (SBSN)-expressing gastric tumor cells activate HSCs into CAFs via EGF/EGFR axis, and cause subsequent productions of CCL2 from HSCs to feed tumor cell growth via CCR2/JAK2 pathway[54]. Sometimes, the HSCs cell activation is not directly driven by tumor cells. Certain immune cells, such as invariant NK cells, can respond to disseminated cancer cells and secrete fibrogenic cytokines such as IL-4 and IL-13, which subsequently trigger Ito cell trans-differentiation into myofibroblasts/CAF-like cells[55].

In summary, multiple tumor-derived factors converge to activate HSCs during metastatic colonization. These include soluble proteins and EV-associated cargo such as AMIGO2, SPHK1, and POU6F2- or SBSN-regulated signaling molecules, which collectively drive HSC transition into myofibroblast-like fibroblasts. Once activated, HSCs remodel the ECM, secrete IL-8 and other chemokines, promote fibrosis, and establish an immune-excluded microenvironment that limits cytotoxic T-cell infiltration and facilitates metastatic outgrowth[52-55].

Hepatocytes

Hepatocytes, which make up nearly 70% of liver mass, are central to the formation of the hepatic metastatic niche through complex interactions with tumor, stromal, and immune cells. Although the precise mechanisms remain incompletely understood, emerging evidence highlights their dual role in modulating immune responses and remodeling the local microenvironment to support metastatic seeding. During early colonization, tumor-secreted FasL can induce apoptosis of neighboring hepatocytes, creating localized areas of parenchymal disruption that facilitate tumor cell invasion[56]. Once in contact with hepatocytes, cancer cells exploit hepatocyte-derived growth and inflammatory factors, including IGF2, hepatocyte growth factor (HGF), fibrinogen-like protein 1 (FGL1), CCL2, and serum amyloid A1/A2 (SAA), to establish a protective and metabolically supportive niche that shelters tumor cells from immune clearance and therapy-induced stress (Figure 1)[57-59].

Activation of the IL-6/signal transducer and activator of transcription 3 (STAT3) pathway in hepatocytes further amplifies this process. STAT3-driven production of SAA recruits myeloid cells and promotes fibrosis, reinforcing an immunosuppressive and fibrotic microenvironment conducive to metastasis[59]. Recent single-cell analyses also reveal a distinct “proinflammatory hepatocyte” subset, primarily derived from periportal zones, that recruits macrophages via the CCL2-CCR2 axis. These recruited macrophages release cytokines such as IL-6, TNF-α, and IL-17, which upregulate PD-L1 on hepatocytes, further suppressing T-cell responses and promoting immune tolerance within the metastatic microenvironment[60]. Beyond cytokine signaling, hepatocytes also exert immunoregulatory control through FGL1, a ligand for lymphocyte activation gene 3. Under physiological conditions, FGL1 maintains hepatic immune tolerance; however, in the metastatic setting, elevated FGL1 inhibits cytotoxic CD8⁺ T cells and NK cells, promoting immune evasion[61].

Collectively, hepatocytes actively sculpt the hepatic niche through inflammatory, metabolic, and immune interactions-transforming from passive parenchymal cells into orchestrators of tumor accommodation and immune suppression.

Angiogenesis

Angiogenesis within liver metastases is a complex, tumor-driven process shaped by both pre-metastatic conditioning and direct tumor-endothelial interactions. Tumor cells secrete proangiogenic mediators such as VEGF, TGF-β, and IL-10 to reprogram local immune and stromal cells, fostering vascular growth and immune evasion[62-64]. In the hepatic metastatic niche, this neovascularization is often aberrant-characterized by disorganized, pericyte-poor, and hyperpermeable (leaky) vessels driven by VEGF-mediated endothelial destabilization and immature vascular architecture[65]. Such dysfunctional vasculature enhances plasma extravasation and stromal remodeling, further facilitating tumor expansion. Angiogenesis in liver metastasis occurs in two coordinated phases: (1) Formation of the PMN, where soluble tumor-derived factors prime the hepatic microenvironment before tumor cell arrival; and (2) Tumor-induced angiogenesis following extravasation. Recent studies have elucidated tumor-type-specific mechanisms driving this process. In colorectal cancer (CRC), hypoxic EVs lacking miR-6086 fuse with hepatic endothelial cells, where hypoxia-induced hypoxia-inducible factor 1α activation suppresses SP1 and thereby downregulates miR-6086 expression. Loss of miR-6086 derepresses angiopoietin-like 4, promoting angiogenesis within the hepatic vasculature[66]. Similarly, CRC-derived exosomes deficient in miR-382-5p fail to inhibit GPR176-GNAS signaling, leading to upregulation of CXCR1/CXCR2 and enhanced vascular permeability and angiogenesis in colorectal liver metastases (CRLM)[67]. In melanoma, thrombospondin-2 expression promotes phenotype switching that enhances vascular invasion and induces neoangiogenesis following hepatic extravasation[68]. Collectively, these findings highlight that liver-specific angiogenesis is shaped by both hypoxia-driven molecular signaling and tumor-derived exosomal communication, with mechanisms varying by tumor type but converging on endothelial remodeling and immune modulation to support metastatic outgrowth.

Other immune populations and coordinated immunosuppression

Beyond KCs, HSCs, LSECs and hepatocytes discussed above, a diverse network of immune cells, including MDSCs, regulatory T cells, M2-like macrophages, IgA⁺ plasma B cells, and neutrophils, contributes to the immunosuppressive landscape of liver metastases. These cells coordinate closely with hepatic stromal elements such as fibroblasts and sinusoidal endothelial cells to suppress cytotoxic lymphocyte activity through cytokine secretion (IL-10, TGF-β), metabolic competition, and checkpoint molecule expression[62,69,70]. NETs and WNT-driven signaling pathways further promote immune exclusion, a hallmark of liver metastases where CD8⁺ T cells remain confined to the invasive margin rather than infiltrating the tumor core[71,72]. Recent preclinical studies have identified WNT11 signaling as a key mediator of this exclusion, promoting IL-17D and suppressing CXCL10 and CCL4 chemokines necessary for T-cell recruitment[73]. While these processes are integral to metastatic immune evasion, many of the underlying mechanisms-such as cytokine and chemokine-mediated suppression, accumulation of immunosuppressive myeloid populations, and exhaustion of cytotoxic T and NK cells-are shared across metastatic sites and have been comprehensively reviewed elsewhere, and thus will not be further elaborated here (Figure 1 and Table 1).

Overall, the coordinated response involving KCs, LSECs, and HSCs-along with a fibrotic, hypoxic, immune-suppressive microenvironment-makes liver metastases uniquely resistant to systemic therapies, including ICIs. Compared to metastases in other organs such as the lung or lymph nodes, the liver metastasis microenvironment is both physically impermeable and immunologically “cold”, resulting in reduced drug penetration, limited T-cell infiltration, and overall poorer therapeutic outcomes[74-76].

ONGOING CLINICAL STUDIES FOR THE MANAGEMENT OF LIVER METASTASES

Many clinical studies have been conducted to improve outcome for those with liver metastasis, and the evidence is summarized below.

Over-arching evidence from meta-analyses

Multiple meta-analyses have consistently shown that liver metastases predict reduced efficacy of ICIs across a range of tumor types. A comprehensive meta-analysis by Tian et al[77] including 163 studies across diverse cancers found that liver metastases were associated with significantly worse overall survival (OS; HR = 1.82, 95%CI: 1.59-2.08) and progression-free survival (PFS; HR = 1.68, 95%CI: 1.49-1.89) in ICI-treated patients (Table 2). In a pan-cancer analysis by Chen et al[78], stratified subgroup analysis of patients receiving ICI monotherapy showed markedly shorter OS for those with liver metastases: 10 months vs 20 months in patients without liver disease (P < 0.0001; Table 1).

Table 2 Clinical evidence summary of immune checkpoint inhibitor outcomes in liver metastases.

Tumor type	Analysis	Treatment/comparison	Endpoint	Liver metastasis subgroup result	Overall population result (for comparison)	Key finding regarding liver metastases
Pan-cancer[77]	Meta-analysis	ICI-treated patients (vs without liver mets)	OS, PFS	OS: HR = 1.82, 95%CI: 1.59-2.08; PFS: HR = 1.68, 95%CI: 1.49-1.89	N/A	Liver metastases associated with significantly worse OS and PFS
Pan-cancer[78]	Pan-cancer analysis	ICI monotherapy	OS	10 months	20 months (patients without liver disease)	Markedly shorter OS for those with liver metastases
NSCLC[81]	Meta-analysis	ICIs vs control	PFS, OS	PFS: HR = 0.64, 95%CI: 0.55-0.75; OS: HR = 0.82, 95%CI: 0.72-0.94	PFS: HR = 0.56, 95%CI: 0.50-0.63; OS: HR = 0.73, 95%CI: 0.66-0.81	Benefit is present but attenuated compared to those without liver mets (larger HR values)
NSCLC[82]	Pooled CheckMate 017/057	Nivolumab vs docetaxel	OS	HR = 0.68, 95%CI: 0.50-0.91	HR = 0.70	Nivolumab maintained an OS benefit, similar to the overall population
NSCLC[83]	KEYNOTE-189	Pembro + chemo vs chemo	OS	HR = 0.62, 95%CI: 0.44-0.87)	HR = 0.56	OS benefit was present but less pronounced than in the overall population
NSCLC[84-87]	Meta-analyses of KEYNOTE-001, KEYNOTE-010, and KEYNOTE-024	ICIs vs control (pooled analysis)	OS	Pooled OS: HR = 0.78, 95%CI: 0.68-0.90	N/A	Survival benefit is present but attenuated relative to patients without liver involvement (ratio of OS-HRs = 1.10)
NSCLC[88]	Real-world cohorts	ICI-treated patients (with vs without liver mets)	OS	N/A	N/A	Approximately 21% higher risk of death (OS: HR = 1.21, 95%CI: 1.17-1.27) with liver metastases
NSCLC[89-91]	IMpower150 (final analysis)	ABCP (atezolizumab + bevacizumab + chemo) vs chemo	OS	HR = 0.68, 95%CI: 0.45-1.02	N/A	Suggested clinically meaningful benefit by adding anti-VEGF (bevacizumab) to the regimen in this subgroup
NSCLC[92,93]	CheckMate 9 LA	Nivo + ipi + chemo vs chemo	OS (5-year rate)	Numerically lower survival than those without hepatic mets; 18% OS rate (vs 11% with chemo alone)	N/A	Durable OS benefit across metastatic subgroups, including those with liver involvement
Melanoma[94]	-	Anti-PD-1 therapy (pooled analyses)	ORR	Markedly reduced	N/A	Liver metastases significantly dampen responses, associated with reduced CD8+ T-cell infiltration
CRC[99]	KEYNOTE-177 (MSI-H/dMMR subset)	Pembrolizumab vs chemotherapy	PFS	Improved PFS (specific HR not provided)	N/A	Pembrolizumab significantly improved PFS, including those with liver metastases
CRC[106]	Real-world, MSI-H/dMMR)	ICI-treated patients (with vs without liver mets)	ORR, PFS	ORR = 58%; PFS: HR = 3.18, 95%CI: 1.52-6.67	ORR 66%	Liver involvement impairs outcomes (lower ORR, higher HR for progression)
CRC[108]	REGONIVO (regorafenib + nivolumab, early-phase)	ICI + VEGF inhibitor	ORR	7%-19% (North American cohorts)	Up to 33% (Japanese cohorts)	Limited generalizability; ORR significantly lower, especially in patients with liver metastases
PDAC[123,125]	KEYNOTE-158 (MSI-H/dMMR subset)	Pembrolizumab	OS, ORR	ORR = 18% (4 of 22 patients); median OS = 4.0 months, 95%CI: 2.1-8.7	N/A	Limited data; liver metastases exacerbate resistance due to fibrosis-driven immune exclusion
Gastric/GEJ cancer[129]	KEYNOTE 859	Chemo-immunotherapy combination	OS	HR = 0.83, 95%CI: 0.70-0.90	HR = 0.73, 95%CI: 0.63-0.84	Survival benefits regardless of liver metastasis status; difference in benefit was not statistically significant
Gastric/GEJ cancer[131]	Retrospective cohort	ICI-treated patients (with vs without liver mets)	OS	10.53 months	13.43 months	No significant difference in OS. Other factors (peritoneal mets, burden) were more strongly associated with reduced PFS
Gastric/GEJ[126]	Mechanistic profiling	-	Immune composition (CD4+, CD8+, monocytic MDSC)	-	Stomach/GEJ show increase proliferating CD4+/CD8+ and decrease monocytic MDSC vs EAC	Proinflammatory TME may blunt LM’s negative effect
Gastric/GEJ[127]	JAVELIN gastric 100 - phase 3 (maintenance)	Avelumab maintenance vs maintenance chemotherapy	OS, PFS	Not reported specifically by LM status	No improvement in OS or PFS with avelumab maintenance	Maintenance ICI alone did not improve outcomes
Gastric/GEJ[128-130]	Meta-analysis	Chemoimmunotherapy vs chemotherapy	OS, PFS	Benefit observed irrespective of LM status across included trials	ChemoIO improved OS and PFS (KEYNOTE859, CheckMate 649)	Supports chemoIO regardless of LM status
Gastric/GEJ[130]	CheckMate 649 - phase 3	Nivolumab + chemo vs chemo	OS, PFS	No dedicated LM subgroup analysis reported	Significant OS/PFS benefit overall (approximately 96% metastatic; approximately 40% LM)	ChemoIO improves outcomes; LMspecific effect not isolated
Gastric/GEJ[129]	KEYNOTE859 - phase 3	Pembrolizumab + chemo vs chemo	OS	HR = 0.83, 95%CI: 0.70-0.90 - benefit regardless of LM	NoLM HR = 0.73, 95%CI: 0.63-0.84	Difference by LM not statistically significant
Gastric/GEJ[131]	Retrospective cohort	Immunotherapy (with LM vs without LM)	OS (factors for PFS)	OS = 10.53 months (LM) - no significant difference vs noLM	OS = 13.43 months (noLM); P = 0.584	Reduced PFS associated with MMR mutations, > 3 metastatic sites, peritoneal metastases

ICI: Immune checkpoint inhibitor; N/A: Not applicable; VEGF: Vascular endothelial growth factor; PD-1: Programmed cell death 1; MSI-H: Microsatellite instability-high; dMMR: Mismatch repair-deficient; MDSC: Myeloid-derived suppressor cell; GEJ: Gastroesophageal junction; EAC: Esophageal adenocarcinoma; NSCLC: Non-small cell lung cancer; CRC: Colorectal cancer; PDAC: Pancreatic ductal adenocarcinoma; OS: Overall survival; PFS: Progression free survival; ORR: Objective response rate; TME: Tumor microenvironment; LM: Liver metastasis.

Together, these meta-analyses confirm that liver metastasis is a robust and independent negative predictive factor for ICI response. These data provide a clear rationale for the investigation and development of liver-specific combination strategies aimed at overcoming resistance to immunotherapy.

Non-small cell lung cancer

Approximately 20% of patients with non-small cell lung cancer (NSCLC) develop liver metastases, which are an established negative prognostic factor and are associated with inferior responses to ICIs[79,80]. Xu et al[81] conducted a systematic review and meta-analysis of 17 randomized controlled trials in NSCLC patients. In those with liver metastases, ICIs improved PFS (HR = 0.64, 95%CI: 0.55-0.75) and OS (HR = 0.82, 95%CI: 0.72-0.94; Table 1). In comparison, patients without liver metastases experienced greater benefits from ICIs, with PFS (HR = 0.56, 95%CI: 0.50-0.63) and OS (HR = 0.73, 95%CI: 0.66-0.81; Table 1)[81].

Standard first-line treatment for advanced NSCLC without actionable driver mutations typically includes PD-1/PD-L1 inhibitors-such as pembrolizumab, nivolumab, or atezolizumab-administered alone or in combination with platinum-based chemotherapy, depending on PD-L1 expression status. Subgroup analysis from the pivotal trials, including KEYNOTE-189 and the pooled CheckMate 017/057 data, consistently demonstrated that the benefit of ICIs is attenuated, but still present, in patients with liver metastases. For instance, in the pooled CheckMate 017/057 analysis, nivolumab maintained an OS benefit over docetaxel in the liver metastasis subgroup (HR = 0.68, 95%CI: 0.50-0.91), which was similar to the HR observed in the overall population (HR = 0.70; Table 1)[82]. Similarly, KEYNOTE-189 showed OS benefit with the combination regimen in patients with liver metastases (HR = 0.62, 95%CI: 0.44-0.87), which was less pronounced than in the overall population (HR = 0.56; Table 1)[83]. Across major pembrolizumab trials, including KEYNOTE-001, KEYNOTE-010, and KEYNOTE-024, patients with hepatic metastases consistently demonstrated lower objective response rate (ORR) and shorter OS compared to those without liver involvement[84-87]. Meta-analyses of these and related studies show that while ICIs improve survival in NSCLC patients with liver metastases (pooled OS: HR = 0.78, 95%CI: 0.68-0.90), the magnitude of benefit is attenuated relative to patients without liver involvement (ratio of OS: HR = 1.10, 95%CI: 0.94-1.29; Table 1)[88]. Real-world cohorts similarly report an approximately 21% higher risk of death among NSCLC patients with liver metastases treated with ICIs vs those without (OS: HR = 1.21, 95%CI: 1.17-1.27; Table 1)[88].

Importantly, IMpower150 demonstrated that adding bevacizumab (anti-VEGF) to atezolizumab and platinum-based chemotherapy (ABCP regimen) improved survival among patients with liver metastases, underscoring the value of vascular normalization in overcoming hepatic immunosuppression[89-91]. The final analysis reported an OS (HR = 0.68, 95%CI: 0.45-1.02) for ABCP vs chemotherapy in the liver metastasis subgroup, suggesting clinically meaningful benefit despite crossing confidence intervals (Tables 2 and 3)[91].

Table 3 Key clinical and translational studies testing combination strategies to enhance immunotherapy in liver metastases.

Strategy	Mechanistic target	Representative trial/evidence	Primary cancer(s)	Outcome summary	Clinical implication
VEGF blockade + ICI[137,138]	Vascular normalization, myeloid suppression	IMpower150 (ABCP regimen)	NSCLC (with liver mets)	OS: HR = 0.68; improved immune infiltration	Supports VEGF + PD-L1 combination
Dual checkpoint blockade[92]	T-cell priming & exhaustion	CheckMate 9 LA	NSCLC, melanoma, CRC	Higher ORR & PFS in liver mets	Increased efficacy but higher toxicity
TGF-β + PD-L1 blockade[158-160]	Fibrosis reversal, stromal remodeling	Bintrafusp alfa trials	Mixed metastatic	Disease control approximately 10%-15%	Promising mechanism; modest efficacy
CSF1R/CCR2 blockade + PD-1[162]	Myeloid reprogramming	Ongoing phase I/II	CRC, NSCLC	Preclinical synergy with PD-1	Pending translation
Epigenetic modulation + ICI[163]	Antigen presentation, viral mimicry	Early-phase trials	Multiple cancers	MHC expression & T-cell infiltration increase	Converts cold to hot lesions
Adoptive cell therapy[166]	Enhanced intrahepatic trafficking	Hepatic artery CAR-T delivery	CRC, HCC	Feasible; improved local homing	Useful for liver-limited disease

ICI: Immune checkpoint inhibitor; VEGF: Vascular endothelial growth factor; PD-1: Programmed cell death 1; PD-L1: Programmed death ligand-1; NSCLC: Non-small cell lung cancer; CRC: Colorectal cancer; OS: Overall survival; PFS: Progression free survival; ORR: Objective response rate; CAR-T: Chimeric antigen receptor-T; HCC: Hepatocellular carcinoma; TGF-β: Transforming growth factor-beta.

Similarly, in CheckMate 9 LA, dual checkpoint blockade with nivolumab plus ipilimumab alongside chemotherapy provided durable OS benefits across metastatic subgroups, including those with liver involvement (Table 3)[92]. At five years, the combination achieved a 18% OS rate vs 11% with chemotherapy alone, with consistent but numerically lower survival in the hepatic metastasis cohort (Table 2)[93]. Together, these findings indicate that although liver metastases confer immunotherapy resistance and attenuated efficacy, the integration of anti-angiogenic or dual checkpoint blockade strategies may partially restore responsiveness and improve outcomes in this challenging subset.

Melanoma

Immunotherapy has revolutionized the treatment of advanced melanoma, with anti-PD-1 monotherapy achieving ORRs of 30%-40% and combination anti-PD-1/CTLA-4 therapy reaching approximately 50% in some settings (Table 2)[94,95]. However, liver metastases significantly dampen these responses[94].

In 2017, Tumeh et al[94] demonstrated that melanoma patients with liver metastases had markedly reduced CD8⁺ T-cell infiltration and experienced inferior responses to anti-PD-1 therapy compared to those without liver involvement. Although the specific ORR numbers in that study represent pooled analyses rather than a single randomized trial, the correlation between poor CD8⁺ infiltration and reduced clinical benefit is robust[94]. There is no dedicated meta-analysis exclusively focused on melanoma liver metastases, but pan-cancer analyses consistently identify liver metastasis as a negative predictive factor for ICI efficacy across tumor types, including melanoma[78].

A recent multi-organ genomic and transcriptomic analysis of therapy-resistant melanoma further substantiates the biological basis of this clinical resistance[96]. In rapid autopsy specimens, liver metastases displayed a distinct immune-desert phenotype characterized by CD8⁺ T-cell depletion, type-2-skewed immunity, and profound T-cell exhaustion[96]. Compared with other metastatic sites, hepatic lesions showed frequent B2M, CDKN2A, and JAK2 loss-alterations associated with impaired antigen presentation and interferon signaling-as well as complement pathway upregulation and suppression of interferon response genes[96]. Ligand-receptor mapping revealed complement (C4A), EPO, and CHEMERIN signaling from adjacent hepatic tissue to tumor cells, and tumor-derived growth hormone signaling to the stroma, highlighting an immunosuppressive, metabolically adaptive hepatic niche[96]. These genomic and microenvironmental features provide mechanistic support for the markedly reduced immunotherapy efficacy observed in melanoma patients with liver metastases.

CRC

Approximately 50%-60% of patients with CRC develop liver metastases during their disease course[97,98]. Standard systemic therapy typically consists of 5-fluorouracil, leucovorin, and oxaliplatin or 5-fluorouracil, leucovorin, and irinotecan, with or without targeted agents. Immunotherapy is reserved for patients with microsatellite instability-high (MSI-H) or mismatch repair-deficient (dMMR) tumors[99,100].

Oligometastasis represents an intermediate state between localized and widespread disease, defined by a limited number of metastatic lesions amenable to curative local therapy[101]. This state is particularly common in CRC because venous drainage from the colon and rectum flows directly into the portal circulation, making the liver the predominant and often the only site of metastasis, found as the sole metastatic site in about one-third of patients at autopsy[102]. Approximately 20%-30% of CRC patients with liver metastases are candidates for resection, which remains the only established curative option, achieving five-year OS rates of 45%-60%, recurrence-free survival of 22%-36%, and long-term cure in 15%-25% of selected patients[103]. Ablative therapies such as stereotactic body radiation therapy, radiofrequency ablation (RFA), cryoablation, and yttrium-90 radioembolization provide effective alternatives for patients who are not surgical candidates, achieving five-year OS rates of 30%-46% and excellent local control for small lesions[104]. The curative potential of liver-only oligometastases is largely unique to CRC, reflecting its slower progression, frequent liver-limited spread, and immune-enriched/desmoplastic microenvironment, features not seen in more aggressive primaries such as pancreatic, lung, or gastric cancers, where liver involvement typically signifies systemic, incurable disease[105].

In KEYNOTE-177, pembrolizumab significantly improved PFS compared to chemotherapy in MSI-H/dMMR metastatic CRC, including those with liver metastases (Table 2)[99]. However, real-world evidence shows that liver involvement still impairs outcomes: In 2024, Saberzadeh-Ardestani et al[106] reported an ORR of 58% with liver metastases vs 66% without, with a HR for disease progression of 3.18 (95%CI: 1.52-6.67; Table 2). For microsatellite stable CRC, which represents the majority of cases, ICI monotherapy has minimal activity, particularly in the presence of liver metastases. This is attributed to the tumor’s immunologically “cold” microenvironment compounded by liver-specific immunosuppression[107]. Combination strategies, including ICIs plus VEGF inhibitors (regorafenib, lenvatinib), MEK inhibitors, radiotherapy, and oncolytic viruses, are under investigation. Early-phase trials such as REGONIVO (regorafenib + nivolumab) reported ORR up to 33% in Japanese cohorts, but North American studies show much lower ORR (7%-19%), especially in patients with liver metastases, indicating limited generalizability (Table 2)[108]. CRC liver metastases are also notable for their dense desmoplastic stroma, which physically impedes T-cell infiltration and drug delivery. This feature, combined with immunosuppressive signaling, makes liver metastases particularly refractory to ICIs even in otherwise immunogenic tumor types[94,97].

Pancreatic ductal adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) exhibits profound desmoplasia that extends to its liver metastases, where metastatic cells reprogram hepatic stellate (Ito) cells into α-SMA⁺ myofibroblasts via TGF-β, PDGF, and FGF2 signaling, driving dense collagen- and fibronectin-rich fibrosis that mirrors the primary tumor’s stroma[109,110]. This fibrosis establishes a physical and immunologic barrier to drug delivery and T-cell infiltration, reinforced by feedback loops between tumor, CAFs, and hepatic fibroblasts, as well as exosome-mediated activation (e.g., CD44v6/C1QBP-IGF1 axis)[111-113]. Single-cell analyses reveal unique stromal subsets (RGS5⁺ CAFs, CTHRC1⁺ fibroblasts) and immunosuppressive SPP1⁺ macrophages sustaining this fibrotic niche[113,114]. Emerging therapies aim to dismantle these barriers: KRAS inhibitors such as MRTX1133 (G12D-selective) and pan-RAS agents (RMC-7977, RMC-6236) remodel the metastatic microenvironment and synergize with ICIs; BiTEs/TriTEs targeting CLDN18.2 enhance T-cell cytotoxicity but are limited by stromal sequestration[115-120]. Combinatorial strategies integrating KRAS blockade, stroma-modulating agents, and immunotherapy, supported by single-cell and spatial profiling, represent the most promising path to overcoming PDAC liver metastasis resistance[111,117]. Standard treatment for metastatic PDAC remains cytotoxic chemotherapy, such as FOLFIRINOX or gemcitabine/nab-paclitaxel[121,122]. ICIs are generally ineffective in PDAC, except in rare MSI-H/dMMR cases[123,124]. In the KEYNOTE-158 trial, 4 of 22 PDAC patients (18%) achieved an objective response to pembrolizumab[123,125]. The median OS was 4.0 months, but the small sample size and wide 95%CI (2.1-8.7 months) limit definitive conclusions (Table 2)[123]. Liver metastases in PDAC exacerbate immunotherapy resistance due to fibrosis-driven immune exclusion, TGF-β-rich microenvironments, and strong recruitment of suppressive immune cells[124]. Combination strategies such as ICIs with chemotherapy, stromal-targeting agents, or TGF-β blockade are being investigated, but no ICI-based regimen has yet demonstrated robust clinical benefit in PDAC patients with liver metastases[124].

Gastric and gastroesophageal junction cancer

Gastric cancer appears to be less dependent on liver metastasis status for its response to immunotherapy than several other tumor types[77]. A proposed mechanism is the presence of a more pro-inflammatory TME in gastric malignancies. Supporting this, Groen-van Schooten et al[126] demonstrated significantly increased proliferating CD4⁺ T helper and CD8⁺ cytotoxic T cells, along with decreased monocytic MDSCs, in tumors located in the stomach and gastroesophageal junction (GEJ) compared with esophageal adenocarcinoma (Table 2)[126].

In gastric cancer, the JAVELIN 100 trial failed to show improvements in OS or PFS with maintenance immunotherapy compared with maintenance chemotherapy (Table 2)[127,128]. Consequently, clinical trial efforts have shifted toward evaluating immunotherapy in combination with chemotherapy, which has shown more promising outcomes.

A meta-analysis by Pu et al[128] of patients with advanced, HER2-negative gastric and GEJ cancers found that combined chemo-immunotherapy improved OS and PFS in both the KEYNOTE-859 and CheckMate 649 trials; Table 2)[129,130]. In CheckMate 649, which enrolled patients with 96% metastatic disease (approximately 40% with liver metastases), combined therapy significantly improved OS and PFS but did not specifically analyze outcomes by liver metastasis status[130]. In KEYNOTE 859, combination therapy produced survival benefits regardless of liver metastasis status. While patients without liver metastases appeared to derive slightly greater benefit, the difference was not statistically significant (HR = 0.73, 95%CI: 0.63-0.84 vs HR = 0.83, 95%CI: 0.70-0.90; Table 2)[129].

A separate meta-analysis by Tian et al[77] further confirmed that liver metastases exert a smaller negative effect on immunotherapy efficacy in gastric and GEJ cancers compared with other tumor types. Instead, peritoneal metastases and the total number of metastatic sites were more strongly correlated with decreased OS. Similarly, a retrospective cohort study by Liang et al[131] reported no significant difference in OS for patients with gastric cancer receiving immunotherapy with or without liver metastases (13.43 months vs 10.53 months; P = 0.584). In this study, MMR mutations (HR = 2.31, 95%CI: 1.09-25.9), metastases involving more than three sites (HR = 2.24, 95%CI: 1.00-5.01), and peritoneal metastases (HR = 3.84, 95%CI: 1.42-10.37) were all associated with reduced PFS (Table 2).

Collectively, these findings indicate that combined chemotherapy and immunotherapy represents a viable treatment option for patients with gastric and GEJ cancers, regardless of liver metastasis status, while other metastatic characteristics, particularly peritoneal involvement and overall metastatic burden, play a more decisive role in determining prognosis.

Strategies to improve immunotherapy efficacy in liver metastases

With evidence that show liver metastases strongly associated with poor response to immunotherapy, several therapeutic strategies have been designed to counteract this resistance by targeting specific pathways and immune barriers.

Targeting neo-angiogenesis: VEGF-directed combinations

A well-established way to modulate the hepatic microenvironment is through VEGF blockade. VEGF drives abnormal, leaky vasculature in liver metastases and recruits immunosuppressive myeloid cells[132-135]. The addition of the anti-VEGF antibody bevacizumab to immunotherapy normalizes vasculature, enhances cytotoxic T cell trafficking and reduces MDSCs and regulatory T-cells (Tregs) (Table 3)[136].

In patients with liver metastases, clinical and translational studies have shown that adding anti-VEGF therapy to ICIs improves immune cell recruitment and disease control compared to ICI monotherapy[137,138]. Clinical studies evaluating ICI plus anti-VEGF therapy in patients with liver metastases, primarily from non-small-cell lung cancer and CRC, suggest that VEGF inhibition may partially restore immune responsiveness by improving T-cell infiltration and reducing myeloid suppression[139]. Although robust pooled data are limited, available subgroup and translational analyses demonstrate higher intratumoral CD8⁺ T-cell density, improved perfusion, and modest gains in PFS compared with ICI monotherapy, supporting the role of vascular normalization as a key mechanism of benefit in the hepatic metastatic niche[89]. However, VEGF inhibition alone is insufficient for durable immune reprogramming[140]. The vasculature tends to renormalize transiently, and compensatory pathways such as angiopoietin-2 and HGF/c-MET signaling can re-establish immunosuppression[141]. Future studies could focus on rational anti-angiogenic combinations that pair VEGF blockade with PD-1/PD-L1 inhibition to sustain vascular normalization, enhance T-cell trafficking, and counteract myeloid-driven immunosuppression within liver metastases (Table 3).

Dual immune checkpoint blockade

Dual checkpoint inhibition targeting CTLA-4 and PD-1/PD-L1 pathways addresses two distinct phases of T-cell dysfunction: Impaired priming in lymphoid tissues and exhaustion within the TME[142]. This strategy is particularly relevant in liver metastases, where effector T-cells are deleted by FasL⁺ macrophages and suppressed by Tregs (Table 3)[45].

Retrospective and prospective analyses across multiple tumor types including melanoma, hepatocellular carcinoma, and CRC demonstrate that dual immune checkpoint blockade (CTLA-4 plus PD-1/PD-L1 inhibition) yields higher ORR and improved PFS in patients with liver metastases compared to monotherapy (Table 2)[143-145]. Mechanistic studies in murine models of liver metastasis show that CTLA-4 blockade preferentially expands tissue-resident effector memory CD4⁺ and CD8⁺ T-cells and reduces intrahepatic Tregs, while PD-1/PD-L1 blockade restores the cytotoxic function of exhausted T-cells within the metastatic niche[144,146]. The combination leads to increased intratumoral infiltration of CD8⁺ and CD4⁺ T-cells, enhanced Th1/M1 cytokine profiles, and a reduction in immunosuppressive myeloid and regulatory populations[147]. These synergistic effects remodel the hepatic immune microenvironment, converting immune-excluded (cold) lesions into inflamed, immune-responsive tumors, thereby improving the efficacy of ICIs in the context of liver metastases (Table 3)[144].

Despite encouraging signals, the benefit of dual ICIs in liver metastases remains modest in magnitude[148,149]. The main challenges are hepatotoxicity and systemic immune-related adverse events, which are more frequent when both checkpoints are inhibited (Table 2)[150,151]. To mitigate toxicity, newer regimens such as “priming” with a single low-dose CTLA-4 antibody followed by PD-1/PD-L1 maintenance are under evaluation in ongoing phase II trials specifically enrolling patients with liver-dominant metastases[152-154].

Targeting TGF-β and myeloid-driven immunosuppression

TGF-β plays a central role in shaping the fibrotic and immune-excluded phenotype of liver metastases[62]. It drives HSC activation, ECM deposition, and conversion of KCs into M2-like macrophages[155]. Therapeutic targeting of this pathway aims to dismantle the fibrotic stroma and restore immune access to the tumor core (Table 3)[156,157].

Dual TGF-β/PD-L1 blockade has shown promising mechanistic results. Agents such as bintrafusp alfa colocalizes a TGF-β “trap” to the TME via PD-L1 binding, resulting in more effective local TGF-β sequestration and simultaneous checkpoint inhibition (Table 2)[158-160]. This approach enhances antitumor immunity in liver metastases by increasing T-cell infiltration, reducing immunosuppressive myeloid activity, and downregulating pathways such as EMT and fibrosis, thereby overcoming resistance to immunotherapy[159,160]. Although clinical responses remain modest, with disease control rates around 10%-15% in heavily pretreated cohorts, these studies validate the biological relevance of TGF-β in liver-specific immune resistance (Table 3)[160].

Alternative approaches include selective inhibition of TGF-β receptor I kinase (e.g., galunisertib) and macrophage reprogramming agents targeting CSF1R or CCR2 signaling to reduce immunosuppressive myeloid infiltration[161]. Preclinical liver metastasis models show that such inhibitors synergize with PD-1 blockade to restore CD8⁺ cytotoxic function and diminish fibrosis. Clinical translation is ongoing, with several phase I/II trials testing these combinations specifically in metastatic colorectal or lung cancers with liver involvement[162].

Epigenetic and cellular reprogramming approaches

Epigenetic modulation represents a frontier strategy to enhance tumor immunogenicity in the liver microenvironment. DNA methyltransferase and histone deacetylase inhibitors can upregulate antigen-presentation machinery (e.g., MHC class I, β2-microglobulin) and reactivate endogenous retroviral elements, leading to “viral mimicry” and interferon signaling (Table 2)[163]. When combined with ICIs, these agents convert immunologically “cold” liver metastases into inflamed lesions with increased T-cell infiltration (Table 3)[164,165].

Similarly, adoptive cell therapies, such as chimeric antigen receptor-T (CAR-T) or T cell receptor-engineered T-cells face unique trafficking and survival barriers within the liver (Table 2)[166]. New strategies utilize chemokine receptor engineering (e.g., CXCR3, CXCR6) to improve intrahepatic homing, and metabolic reprogramming to resist TGF-β-mediated exhaustion. Trials delivering CAR-T-cells intra-arterially via the hepatic artery are underway to bypass the sinusoidal barrier and achieve higher local concentrations (Table 2)[167,168].

Ablative approaches for the management of liver metastases

Ablative therapies-including RFA, microwave ablation (MWA), cryoablation, and stereotactic body radiotherapy (SBRT)-offer effective local control for liver metastases across multiple primary cancers, with particularly strong evidence in CRC (Table 4)[169]. For small, oligometastatic CRLM, ablation can be potentially curative, achieving local control rates of 80%-95% and OS comparable to surgical resection (Table 3)[170]. The phase III COLLISION trial confirmed non-inferiority of ablation vs surgery for OS (HR = 1.05, P = 0.83) and local control, while the CLOCC trial demonstrated a durable OS benefit when RFA was added to systemic therapy in unresectable CRLM (8-year OS 35.9% vs 8.9%, HR = 0.58, P = 0.01; Table 3)[171,172]. For other primaries (breast, lung, neuroendocrine), ablation provides high local control (≥ 85%) and symptom palliation, though no randomized data confirm an OS benefit[173]. SBRT achieves local control rates of 83%-94% for small lesions (< 3 cm) with minimal grade ≥ 3 toxicity (< 10%), and MWA may outperform RFA for perivascular tumors due to reduced heat-sink effects (Table 3)[174,175]. Emerging strategies combine ablation with ICIs to augment systemic immunity[176]. In hepatocellular carcinoma, IMbrave050 and Qin et al[177] and Zhu et al[178] trials showed improved recurrence-free or OS with ablation + ICI compared to ICI alone, though OS benefit remains unconfirmed in long-term follow-up (Table 3). In contrast, the EORTC-1560 ILOC trial in CRC failed to demonstrate responses in untreated metastases, suggesting liver-specific immune tolerance limits efficacy[179]. Overall, ablation offers excellent local disease control and safety across cancer types, potential cure for limited CRLM, and serves as a promising immunomodulatory adjunct to ICI in hepatocellular and possibly metastatic settings, warranting further biomarker-driven trials[180].

Table 4 Summary of clinical evidence for ablative therapies in liver metastases across tumor types.

Technique	Typical local control rate	Any OS benefit in RCTs	Comments
RFA[169]	80%-95% (CRLM)	Yes (CLOCC: 8-year OS 35.9% vs 8.9%, HR = 0.58, P = 0.01)	Effective for small, oligometastatic CRLM; potentially curative; added to systemic therapy in unresectable CRLM
MWA[174,175]	≥ 85% (various primaries)	Not reported	May outperform RFA for perivascular tumors due to reduced heat-sink effects
Cryoablation[169]	≥ 85% (various primaries)	Not reported	High local control for breast, lung, neuroendocrine metastases; symptom palliation
SBRT[169]	83%-94% (small lesions < 3 cm)	Not reported	Minimal grade ≥ 3 toxicity (< 10%)
Ablation (general)[171,172]	80%-95% (CRLM)	Yes (collision: HR = 1.05, P = 0.83, non-inferior to surgery)	Non-inferior to surgery for OS and local control in CRLM; promising with ICI
Ablation + ICI[177,178]	Not specified	Yes (IMbrave050: Improved RFS/OS vs ICI alone)	Improved RFS/OS in HCC; unconfirmed long-term OS benefit; EORTC-1560 ILOC showed no response in untreated metastases

ICI: Immune checkpoint inhibitor; OS: Overall survival; RFS: Recurrence-free survival; HCC: Hepatocellular carcinoma; RFA: Radiofrequency ablation; CRLM: Colorectal liver metastases; RCT: Randomised controlled trial; MWA: Microwave ablation; SBRT: Stereotactic body radiotherapy.

Translational challenges and future directions

Despite encouraging mechanistic progress, the clinical translation of these strategies faces several barriers. First, biomarker validation remains limited: Traditional markers such as PD-L1 expression or tumor mutational burden fail to predict ICI response in hepatic metastases[45,94]. Emerging tissue, and blood-based biomarkers such as CD8⁺ T-cell density, interferon-γ-related gene signatures, and post-treatment reduction in exhausted T-cell subsets (PD-1⁺ TIM-3⁺, TIGIT⁺) show promise but require standardization[181].

Second, trial design complexity impedes progress. Most ICI studies stratify by primary tumor type rather than by metastatic organ site, diluting the statistical power to detect liver-specific effects. Dedicated liver-metastasis cohorts or platform trials incorporating multi-omic profiling could clarify which patients benefit from combination strategies.

Finally, safety and tolerability pose practical challenges. The liver’s dual role as a metabolic and immune organ heightens the risk of immune-related hepatotoxicity, particularly with dual checkpoint or TGF-β-targeting regimens[182-184]. Rational sequencing, such as using VEGF or radiation “priming” before full systemic immunotherapy may help mitigate toxicity while maximizing efficacy[185].

CONCLUSION

Liver metastases represent a biologically distinct and clinically formidable barrier to effective immunotherapy. Their resistance arises not from tumor-intrinsic factors alone, but from the liver’s unique immune architecture, which favors tolerance over inflammation. Throughout this review, we have highlighted how hepatic-specific features such as sinusoidal structure, KC-mediated T-cell deletion, and stellate cell-driven fibrosis collectively generates a profoundly immunosuppressive niche that limits the efficacy of ICIs across multiple tumor types. Mechanistic and clinical evidence converge on a key insight: The liver does not merely host metastases but actively reprograms immune dynamics systemically. This realization has reframed therapeutic development from conventional ICI monotherapy toward strategies that remodel the hepatic microenvironment itself. Approaches such as VEGF inhibition, dual checkpoint blockade, and liver-directed radiation aim to normalize vasculature, restore cytotoxic T-cell trafficking, and reduce myeloid-mediated suppression. Meanwhile, next-generation interventions like targeting TGF-β signaling, reprogramming suppressive macrophages, and employing epigenetic or adoptive cellular therapies are redefining how immune exclusion in liver metastases can be reversed. Despite promising preclinical data, clinical translation remains limited by heterogeneous trial designs, lack of validated biomarkers, and concerns over hepatotoxicity. The path forward will require mechanism-guided trials specifically stratified by hepatic involvement, integration of multi-omic profiling to capture immune and stromal evolution, and rational sequencing of locoregional and systemic therapies. Ultimately, overcoming immunotherapy resistance in liver metastases demands a paradigm shift from treating the liver as a passive site of disease to understanding it as an active immune organ that must itself be therapeutically reconditioned. Success will depend on the ability to integrate immunologic, vascular, and stromal modulation into unified treatment strategies that transform the hepatic niche from a site of immune privilege control.