Stromal cells are actively involved in remodeling the ECM through the production and secretion of components like collagen and fibronectin. Although cancer-associated fibroblasts (CAFs) are the major stromal population, activated hepatic stellate cells (HSCs) are also important contributors to the creation of pro-tumoral immunosuppressive niche in the liver and a great source of CAFs [23,24]. Moreover, they release growth factors and cytokines, such as, CXCL5, TGFβ, and VEGF which favor inflammation, immune escape and angiogenesis, respectively, shaping tumor growth, invasion and metastasis and, thus, the patient’s prognosis [25–27].

Tumor-associated endothelial cells (ECs) play a key role in angiogenesis, however the formed vasculature within the metastatic tissue is abnormal and “leaky”, which affects the amount of nutrients and oxygen that are delivered to tumor cells, contributing to the formation of hypoxic regions. Not only can ECs constitute a barrier for immune cell trafficking, but they can also modulate their activity or metabolism—specially T cells—by expressing certain molecules (PD-L1, FasL...) and/or enzymes, like indoleamine 2,3-dioxygenase 1 (IDO1) or arginase (ARG1) [28].

Another key point in TME is the involvement of the immune system, characterized by many different cell populations among tumor-infiltrating lymphocytes (TILs)—i.e., T cells, B cells, NK cells and myeloid-derived cells—e.g., tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), and dendritic cells (DCs) [29].

Tumor-Infiltrating Lymphocytes (TILs)

Within the T cell compartment, multiple functionally distinct subsets can be distinguished. Cytotoxic T lymphocytes (CD8⁺) play a central role in mediating anti-tumor immunity, releasing factors like granzyme B, perforin, and interferon γ (IFN-γ) to eliminate tumor cells [30]. In fact, high infiltration was associated with better prognosis in CRC patients [31]. However, these cells are often exhausted due to adverse conditions like high reactive oxygen species (ROS) and long-term antigen exposure in the TME. This dysfunctional state is marked by reduced cytokine production and high expression of immune checkpoints such as PD-1 [10,32]. T helper cells (Th, CD4⁺) regulate both innate and adaptive immune responses and consist of multiple subsets whose differentiation and function are shaped by the microenvironment cues. Certain subsets support anti-tumor immunity; for example, Th1 cells, which secrete IFN-γ, tumor necrosis factor alpha (TNF-α), and interleukin 2 (IL-2), or cytotoxic CD4⁺ T cells, which can directly eliminate tumor cells expressing MHC class II [33]. In contrast, subsets such as Th9 (IL-9 producers) and Th17 (IL-17A producers) exhibit context-dependent roles: while they can promote tumor progression by enhancing angiogenesis and suppressing immune responses, they may also facilitate tumor rejection by recruiting and activating cytotoxic immune cells. Th22 cells, characterized by IL-22 expression, contribute to an immunosuppressive microenvironment, partly through the induction of IL-10 production [34]. Regulatory T Cells (Tregs)—characterized by expression of FOXP3 and CD25—suppress immune response by releasing inhibitory factors (IL-10, TGF-β and IL-35) and maintaining immune checkpoints, which reduce cytotoxic T cells activity and inhibit dendritic cells maturation [35]. Their accumulation was associated with poor prognosis in many tumor types, including metastatic CRC [22,36].

B cells are known for presenting antigens and producing antibodies, which can help kill malignant cells and, thus, improve patients’ prognosis. However, a subset called regulatory B (Bregs) cells can enhance angiogenesis, reduce cytotoxic T cell activity by expressing PD-L1 and immunosuppressive cytokines such as IL-10 and TGF-β and, thus, foster immune evasion [17,37].

Natural killer (NK) cells are potent cytotoxic effectors and part of the innate immune system. They can eliminate tumor cells regardless of antigen presentation by producing granzyme B and perforin or facilitating programmed cell death through expression of FasL and TNF-related apoptosis-inducing ligand (TRAIL) [38]. Nonetheless, NK cells can present an exhausted phenotype, characterized by PD-1 expression and impaired function [39].

Myeloid-Derived Cells

TAMs are the most abundant infiltrating immune cells in the TME, often accounting for a substantial proportion—up to 50%—of the total cellular content [40]. Macrophages have elevated plasticity and can adopt diverse phenotypes in response to environmental factors. While M1 phenotype favors inflammation and exhibit anti-tumor activity, M2 supports disease progression due to its immunosuppressive nature. They secrete factors like arginine, IL-10, transforming growth factor-beta (TGF-β) or C-C motif chemokine ligand 5 (CCL5) that inhibit cytotoxic T cell activity and recruit Tregs- and promote ECM remodeling and angiogenesis. In the liver TME, TAMs showed different morphology and localization, which has been correlated with patient prognosis, the poorest outcome in those patients with predominant bigger and pro-tumorigenic M2-like [41–43]. Their presence has also correlated with poor prognosis [22,44,45]. In addition, Kupffer cells, the liver-resident macrophages, can support tumor immune evasion by expressing immune checkpoint molecules which suppress CD8⁺ T cell effector function [46]. Highly metastatic CRC cells potentiate the polarization of Kupffer cells toward an M2 phenotype, thereby promoting tumor invasion and metastasis [47]. Moreover, tumor-educated Kupffer cells actively contribute to immune tolerance by facilitating Tregs priming [3] and promoting the expansion of MDSCs through the secretion of IL-10 and TGF-β [48–50].

MDSCs are immature myeloid progenitors, often enriched in the metastatic liver, which foster an immunosuppressive environment that supports tumor cell growth and immune escape [51,52]. They can inhibit T and NK cells’ function by direct contact killing and/or depletion of essential amino acids crucial for their viability and they can promote Tregs cell expansion through releasing factors like arginine, ROS and immunosuppressive cytokines [22,53,54].

DCs are specialized antigen-presenting cells (APCs) that capture, process, and present antigens to T cells, thus playing a central role bridging innate and adaptive immunity, orchestrating effective immune surveillance against cancer [55]. Infiltration of mature DCs within the tumor has been correlated with better prognosis in multiple cancer types including CRC [56], however, their function is frequently impaired within the TME of liver metastases, limiting anti-tumor immunity [52].

Granulocytes compile neutrophils, eosinophils, and basophils. Among them, tumor-associated neutrophils (TANs) have attracted more attention in oncology as they can be polarized in response to environmental factors such as TAMs, acquiring pro or anti-tumor activity and influencing different processes like angiogenesis, matrix remodeling, and invasion through signaling molecules [52]. Higher infiltration of TANs was found in the peritumoral area of CRC tumors with mutated KRAS, in comparison with wild-type KRAS [57].

Mast cells (MCs) trigger inflammatory responses through the release of histamine, proteases, cytokines (IL-6, IL-9, IL-13 and TNF) and chemokines (CXCL8, CCL2 and CCL5) when activated. They can modulate angiogenesis, vascular permeability, ECM remodeling and tumor invasion [22]. Indeed, low levels of MCs were associated with better outcomes in CRC patients through decreased vascularity [11].

Given the mechanism of action of ICIs, the presence of a sufficiently high TMB along with right antigen presentation is a key determinant of response to PD-1/PD-L1 blockade [98,99].

TMB is typically low (<10 mutations/Mb) in MSS CRC resulting in poor immunogenicity and reduced neoantigen-driven CD8⁺ T cell activation, which contributes to treatment resistance [100].

Moreover, impaired antigen presentation profoundly impacts immune recognition. Loss or downregulation of MHC class I molecules may occur via β₂-microglobulin (β2m) mutations, epigenetic silencing of HLA genes, or deficiencies in antigen-processing components such as TAP1, tapasin, and LMP2/TAP2. For instance, TAP expression is reduced in KRAS-mutated CRC, which compromises peptide presentation to cytotoxic T lymphocytes [80,101]. β2m loss and MHC-I dysfunction can even occur in high MSI CRC with robust T cell infiltration due to immune selection [101].

In the context of liver metastases, antigen presentation may be further impaired by the tolerogenic hepatic microenvironment, which promotes antigen uptake by Kupffer cells and liver sinusoidal endothelial cells, leading to T cell inactivation [80].

Immune resistance can be also mediated by several oncogenic pathways intrinsic to tumor cells, through suppressing T cell infiltration and regulating PD-L1 expression also affected by inflammatory cues.

Aberrant activation of the WNT/β catenin pathway signaling correlates with a reduced CD8⁺ T cell infiltration in CRC as it decreases the expression of recruiting chemokines such as CCL4 while it increases immune checkpoints like CD47 [102–104]. MAPK pathway activation—secondary to RAS mutations—enhances PD-L1 expression by mRNA stabilization [80]. Moreover, loss of PTEN activates the PI3K/AKT axis, which leads to PD-L1 upregulation and creation of a “cold” TME in multiple tumor types, including CRC [105].

Transcription factors such as STAT3 contribute to the upregulation of PD-L1 on tumor cells, both through intrinsic signaling pathways and in response to external stimuli. Proinflammatory cytokines like IFN-γ and IL-6 enhance this effect via activation of the JAK/STAT1 and STAT3 pathways, respectively. However, defects in IFN-γ signaling, arising from mutations in JAK1/2, IFN-γ receptors, or downstream effectors like IRF1, impair the induction of MHC molecules and antigen presentation machinery, ultimately diminishing the effectiveness of PD-1/PD-L1 blockade [106].

Although over 50% of CRC patients may exhibit tumor PD-L1 positivity (≥10% cutoff), PD-L1 alone remains an unreliable biomarker for predicting ICI response due to challenges such as intratumoral heterogeneity, discordance between primary and metastatic lesions, and a lack of standardized scoring criteria [11,107].

The reduced presence or dysfunction of T cells within the TME represents a critical factor influencing immunotherapy response. Even when antigens are present, the immune system’s capacity to reach an effective response is hindered by the limited infiltration or senescence of effector T cells [108]. This is particularly challenging in CRC, as MSS tumors typically exhibit a non-inflamed or “cold” TME characterized by low densities of TILs, especially CD8⁺ cytotoxic T cells, compared to high MSI tumors. Many CD8⁺ T cells in MSS CRC lack CD28 expression, a critical co-stimulatory molecule, suggesting a senescent-like state with reduced proliferative and cytotoxic capabilities [109].

Additionally, in liver metastases, this T cell dysfunction is exacerbated. Clinical and preclinical studies have shown that the hepatic niche is particularly refractory to ICIs due to reduced marginal CD8⁺ T cell infiltration, and accumulation of PD-1⁺ TAMs at the TME of metastases [19]. Notably, metastatic liver disease has been shown to induce systemic deletion or exhaustion of tumor-specific CD8⁺ T cells, impairing immunotherapy efficacy even at distant tumor sites [96,110,111].

Besides the immune checkpoints targeted with FDA/EMA-approved ICIs, there are other factors that can cause T-cell exhaustion and modify the patients’ immune response. Indeed, PD-1/PD-L1 blockade efficacy can be undermined in MSS CRC tumors when other inhibitory pathways, including CTLA-4, TIM-3, LAG-3, TIGIT, VISTA, and IDO1, are also activated [80]. While LAG-3, expressed on activated T cells, Tregs, and NK cells, suppresses immune activation by binding MHC class II molecules [112], TIGIT, expressed on T cells and NK cells, binds to CD155 and CD112 on tumor cells or APCs and competes with the co-stimulatory receptor CD226, exerting its inhibitory effect on immune activation [113]. These are frequently simultaneously expressed on exhausted CD8⁺ and CD4⁺ T cells, as well as on Tregs, reinforcing a suppressive immune environment. For instance, the presence of LAG-3⁺ FOXP3⁺ Tregs in CRC has been associated with poor clinical outcomes, and the co-expression of PD-1 with CTLA-4 or LAG-3 on CD8⁺ T cells is indicative of deeper functional exhaustion, often refractory to immunotherapy [114,115].

Liver metastases represent a particularly hostile immune environment, characterized by a unique immunosuppressive milieu which often leads to therapeutic resistance even in tumors with immunogenic features such as high MSI status. Among the challenging factors that promote immune evasion are a marked accumulation of PD L1⁺ TAMs, expansion of MDSCs, and enrichment of Tregs, working together to create a profoundly immunosuppressive TME that hampers cytotoxic T cell infiltration and function, thereby limiting the efficacy of PD-1/PD-L1 blockade [104].

TAMs in CRC are predominantly inclined toward an M2-like phenotype, while single-cell studies have identified a particularly enriched population of SPP1⁺ M2-like macrophages, which induce T cell apoptosis through IL-10/STAT3 signaling and Fas/FasL mechanisms and whose presence is associated with poor prognosis [104]. Furthermore, TAMs contribute to immune evasion by secreting immunosuppressive cytokines such as IL-10, TGF-β, and pro-angiogenic factors such as VEGF and by releasing chemokines like MCP 1 and MIP 1α/β that facilitate tumor cell invasion and further immune suppression [80,101,116]. Besides, PD-L1 expression is frequently observed not only in tumor cells but also in surrounding TAMs in liver metastases, contributing to T cell suppression through the PD-1/PD-L1 axis and correlating with resistance to immunotherapy and poorer outcomes [80,117–119]. Interestingly, not only the expression profile but also the morphology of TAMs at the interface of CRC liver metastases correlates with patient survival and has been shown to predict disease recurrence after hepatectomy [42].

MDSCs, particularly polymorphonuclear (PMN-MDSCs) and monocytic (M-MDSCs) subtypes, expand in the metastatic TME, especially under IL 1β–rich conditions [86]. Importantly, this immunosuppressive phenotype is already present in the hepatic pre-metastatic niche and has been clearly associated with tumor progression and invasion. These cells impair T cell function or even deplete them by inducing apoptosis while they also promote Treg differentiation, hindering the antitumor immune response [116,120].

Tregs are highly enriched in CRC compared to healthy tissue and play a critical role in immune evasion. They contribute to immune suppression through various mechanisms. For one, they secrete immunosuppressive cytokines such as IL-10, TGF-β, and IL-35, consume IL-2, depriving effector T cells from this crucial growth factor, and, for another, they can directly induce apoptosis in both APCs and cytotoxic T cells via granzyme B and perforin pathways [116].

In addition to cellular components, the TME is enriched with soluble factors and metabolites that further hold immune activity back.

Among the most potent of these mediators is TGF-β, secreted by CAFs and TAMs, which promotes Treg induction and drives immune exclusion, particularly in the mesenchymal CMS4 subtype of CRC [104]. IL-10, secreted by both Tregs and TAMs, hampers CD8⁺ T cell activity by inhibiting T cell receptor signaling and downregulating key effector molecules [116]. IL-17A has been shown to contribute to the creation of a protumor environment by promoting MDSC recruitment and PD-L1 upregulation, while IL-17A blockade could restore sensitivity to PD-1 inhibition in preclinical CRC models by enhancing cytotoxic T cell infiltration [121]. Additionally, elevated serum levels of IL-22 have been associated with chemoresistance in CRC patients, suggesting its potential involvement in broader immune escape mechanisms [34]. Moreover, additional immune checkpoints such as IDO, VISTA, and CD73 can be found within the TME [122], which further reinforces immune resistance.

The presence of specific metabolic byproducts within the TME can also suppress T cell function. In addition to competition with tumor and stromal cells for essential nutrients such as glucose, glutamine, and tryptophan, this metabolic stress results in nutrient-deprived or “starved” T cells, leading to an immunometabolic crisis that impairs their survival and effector functions, including altered cytokine production. For instance, dysregulated glutamine metabolism in TME has been correlated with poor immune infiltration, contributing to immune evasion and resistance to ICIs in CRC tumors [123]. Furthermore, accumulation of lactate and kynurenine further dampens T cell activity [116]. Extracellular adenosine, generated by stimulation of CD39- and CD73-expressing Tregs and TAMs, binds to A2A receptors on T cells and suppresses their function [124,125].

Chemotherapy can restore immune surveillance even in immune-depressed tumors by inducing cell death, releasing tumor antigens and danger signals, which enhances DCs activation and T-cell priming [22,128,129]. Indeed, adding ICIs to conventional antitumor drugs (i.e., oxaliplatin) has shown enhanced anti-tumor responses in preclinical models [130] and in the subset of MSS metastasized CRC in randomised METIMMOX phase II trial (NCT03388190) [131]. Similarly, clinical studies have demonstrated the benefit of combining anti-PD-1 therapy with anti-angiogenic such as bevacizumab [132] or tyrosine kinase inhibitors (TKIs) such as regorafenib, as they can normalize tumor vasculature which, in turn, relieves hypoxia and allowing greater T-cell infiltration and a reduction of immunosuppressive populations [126]. Besides, regorafenib has been shown to suppress MDSC and reprogram DCs’ function, sensitizing CRC models to ICIs [126] and CRC patients with advanced MSS CRC without liver metastases in a phase I non-randomized City of Hope Study (NCT04362839) [133]. Notably, the addition of atezolizumab, a PD-L1 inhibitor, to a first-line chemotherapy regimen of FOLFOXIRI with bevacizumab reported survival benefits in both MSS and high MSI CRC in the ongoing AtezoTRIBE trial (NCT03721653), attributed in part to enhanced antigen release and vessel remodeling, allowing greater T-cell infiltration into tumors [126,134,135].

In addition, alternative pathways involved in immune evasion are being investigated as therapeutic targets to improve responses to immune checkpoint inhibitors. For instance, pro-tumorigenic MAPK signaling, commonly activated in CRC, can lead to immune resistance by impairing T cell infiltration and promoting immunosuppressive gene expression. In this context, MEK inhibitors have been used to block MAPK signaling, which in turn preserved CD8⁺ T cell function and sensitized tumors to PD-L1 blockade in preclinical CRC models [136] and in BRAF-mutated CRC patients within a clinical trial (NCT03668431) [137]. Furthermore, the inhibition of cyclin-dependent kinases 4 and 6 (CDK4/6) with small molecules has proven to increase the response to PD-1 blockade by rising tumor infiltration and activation of effector T cells [138]. Besides, the potential benefit of combining ICIs with inhibitors of immunosuppressive signaling pathways, such as IDO, TGF-β or WNT/β-catenin, is also being explored in ongoing trials to potentiate immune responses and achieve better treatment responses [139,140].

Another relevant mechanism is autophagy, a cellular process essential for maintaining homeostasis, which also shapes the tumor immune microenvironment. Modulation of autophagy has been shown to enhance antigen processing and presentation, leading to improved recognition by cytotoxic T cells. Therefore, pharmacological inhibitors or modulators of autophagy have been combined with anti-PD-1/PD-L1 therapy, showing promising results in both in vitro and in vivo tumor models, including enhanced CD8⁺ T cell infiltration and tumor regression [141,142].

Liver-directed radiotherapy (RT) can reshape the immunosuppressive liver milieu. Thus, RT combination with ICIs has emerged as a promising strategy to enhance treatment efficacy in liver metastasized CRC [143]. In animal models, liver RT restored systemic antitumor immunity by depleting immunosuppressive TAMs and promoting T cell survival and infiltration when combined with ICIs [111].

The triple combination of RT, anti-PD-1 and anlotinib (anti-angiogenic TKI) was tested in vivo, using a mouse liver metastasis tumor model, to achieve a synergistic effect by acting on different targets: locoregional therapies induce cell death and antigen release, checkpoint blockade reactivates T-cell effector functions and anti-angiogenic agents normalize the vasculature. A significantly higher infiltration of CD4⁺, CD8⁺ T cells and less MDSCs were observed in treated mice, which translated into improved disease control. Together, this combination remodeled the liver TME from immunologically “cold” to “hot,” thereby enabling effective antitumor immune responses [144]. Promising results were also observed when combining anlotinib and PD-1 blockers in patients with resistant CRC metastasis [145].

Another triple combination—envafolimab (anti-PD-L1), with short-course RT and CAPEOX-based chemotherapy—was tested on locally advanced rectal cancer within the PRECAM trial (NCT05216653), which resulted in a pathologic complete response rate of 62.5% among MSS patients [126,146].

Targeting more than one immune checkpoint at once could help overcome T-cell exhaustion (i.e., combining anti-PD-1 + anti-CTLA-4 agents) [80]. In the NEST trial (NCT05608044), botensilimab (anti–CTLA-4) and balstilimab (anti–PD-1) achieved durable immunologic responses when used as a neoadjuvant regimen in patients with localized MSS CRC [126]. Furthermore, dual immune checkpoint blockade with durvalumab and tremelimumab (anti-PD-L1 and anti-CTLA-4, respectively) was tested in combination with mFOLFOX6 chemotherapy within the single-arm MEDITREME phase 1b/2 trial (NCT03202758), demonstrating high disease control in RAS-mutated MSS metastatic CRC [129].

Emerging immune checkpoints upregulated in dysfunctional T cells within the TME such as LAG-3, TIGIT, TIM-3, and VISTA are being explored in combination with PD-1/PD-L1 inhibitors aiming to restore effector function and broaden the therapeutic response, particularly in MSS and liver-metastatic CRC, where single-agent immunotherapy has limited efficacy [147–149]. For example, dual PD-1/LAG-3 blockade (pembrolizumab plus favezelimab) has shown synergistic effects by reaching modest clinical responses in MSS CRC within an early-phase clinical trial [149,150]. Similarly, tiragolumab (anti-TIGIT) in combination with atezolizumab (anti-PD-L1) have demonstrated promising efficacy in solid tumors, including high MSI CRC [113,149].

Beyond ICIs’ combination, bispecific antibodies (BsAbs), which bind two different antigens or epitopes, represent an innovative strategy to simultaneously engage immune cells and tumor targets. A PD-L1/CD3 BsAb was designed to activate CD3 on T cells while helping them link to PD-L1⁺ tumor cells and demonstrating potent antitumor activity in preclinical CRC models. This effect was further enhanced when combined with regorafenib [151,152].

Although combining ICIs with other agents had shown promising response rates, it is also associated with higher frequency and severity of immune-related toxicities compared to single-agent treatments. Therefore, careful patient selection and close monitoring of side effects are critical when implementing these strategies [96,153].

Different therapies to improve antigen presentation and increase tumor “visibility” to the immune system are under investigation, such as restoring MHC-I expression through epigenetic modulation or stimulating antigen-presenting cells [101,116].

Neoantigen-based vaccines could potentially boost tumor recognition and improve survival for patients in immune-cold subtypes. Thus, cancer vaccines and oncolytic viruses are under development to stimulate endogenous DC activation and boost T cell priming [126]. For instance, machine learning was used to identify tumor antigens and immune targets and tailored to immune subtype-specific responses for CRC patients [152].

Other strategies to boost DC function and overcome tolerogenic signaling include in vivo DC activation using GM-CSF or their ex vivo expansion, maturation and reinfusion in the patient [163].

The ECM and stromal cells in liver metastases can physically and biochemically exclude immune cells. Hence, strategies that interfere with stromal barriers, such as targeting CAFs or HSCs, or that modulate ECM components (i.e., collagenases), could facilitate T cell infiltration and improve therapeutic efficacy [164]. Interestingly, inhibiting activated HSCs through up-regulating the expression of the relaxin (RLN) resolved liver fibrosis and had anti-metastasis effect with PD-L1 in liver metastasized CRC mouse models [165]. However, a human monoclonal anti-fibroblast activation protein (anti-FAP) antibody showed no clinical activity in CRC patients with liver metastasis within a phase I trial [22].

Hypoxia, common in liver metastases, also drives immunosuppression; novel methods such as manganese-based nanoparticles combined with small-molecule modulators have been proposed to locally increase ROS and provide antitumor treatment and improve the efficacy of immunotherapy [166].

Tumor-derived and stromal cytokines can recruit immunosuppressive cells. Therefore, inhibiting pro-tumor cytokine is a strategy to tilt the balance toward immunity. For instance, preclinical work shows that inhibiting IL-17A or IL-1β can blunt MDSC recruitment and reinvigorate CD8⁺ T cells in CRC models [80,86].

NK cells could be especially useful in immunologically “cold” tumors like MSS CRC due to their ability to recognize “dangerous” cells independently of MHCs and secrete cytokines—such as IFN-γ, IL-4, and TNF-α—which promote cytotoxicity, enhance DC function, and help recruit CD8⁺ T cells. Strategies using NK cell agonists, α-GalCer–loaded DCs or synthetic glycolipid vaccines are being investigated to activate both NK and CD8⁺ T cells and synergize with DC-based therapies [163].

Adoptive cell therapies, particularly CAR-T cell approaches, are being adapted to solid tumors like CRC to bypass immune evasion mechanisms and directly engage tumor-specific antigens. In liver metastases of CRC, preclinical studies using CAR-T cells targeting cadherin-17 (CDH17), a cell adhesion molecule highly expressed in CRC but largely absent from normal tissue, have shown potent antitumor activity. Systemically administered CDH17-targeted CAR-T cells significantly reduced tumor burden in humanized mouse models [167]. However, translating CAR-T therapy into CRLM faces unique challenges imposed by the tolerogenic hepatic environment. Liver sinusoids can trap and deactivate CAR-T cells through phagocytosis (mediated by TAMs or Kupffer cell) and immunosuppressive molecules such as IL-10, IDO and adenosine. Moreover, antigen heterogeneity between primary and metastatic lesions predisposes to antigen escape following treatment [168,169]. To overcome these barriers, multiple optimization strategies are under investigation. These include regional delivery approaches, such as hepatic artery infusion to enhance intratumoral CAR-T accumulation; dual-target CAR constructs designed to reduce antigen-escape variants [170]; and “armored” CAR-T cells engineered to resist or remodel the suppressive niche, for example by incorporating TGF-β-resistant signaling domains [171] or by secreting IL-7 and CCL19 to improve local T-cell recruitment and survival [172,173].

Targeting specific immunosuppressive cell populations within TME represents another promising strategy to enhance the efficacy of immune ICIs in metastatic CRC.

Agents such as CSF1R inhibitors can deplete TAMs or reprogram them toward a pro-inflammatory M1 phenotype, restoring antitumor cytokine production such as IFNγ. Additionally, CCR5 blockade, which disrupts the CCL5-CCR5 axis active in both tumor and immune cells, has shown potential to reverse pro-metastatic inflammation and repolarize macrophages in patient-derived CRC models. These findings were supported by early-phase clinical trials demonstrating clinical responses in patients with advanced CRC liver metastases [117,174].

Furthermore, approaches aimed at limiting MDSC recruitment (via CCR2/CCR5 or CXCR2 inhibitors), reprogramming their suppressive phenotype (e.g., with PI3Kγ inhibitors), or depleting them altogether have shown encouraging results in preclinical models [54,175].

Therapeutic strategies to reduce Treg activity or rebalance the Treg/Teffector ratio include TLR agonists, anti-CD25 antibodies, cyclophosphamide, and IDO inhibitors [117].