P-glycoprotein (P-gp) can be transferred between cells via two pathways independent of drug selection pressure. P-gp can be transferred remotely by the release of particles, which can be effectively detected within 4 h of release. In contrast, contact transfer is mediated by tunneling nanotubes that form “bridges” between neighboring cells (37). For example, paclitaxel-resistant A2780/PTX ovarian cancer cells secrete P-gp-carrying microvesicles (MVs). These microvesicles transferred P-gp into chemotherapy-sensitive A2780/WT cells in a time- and dose-dependent manner, significantly enhancing the latter’s resistance (38). Moreover, the ABCB1 gene encodes P-glycoprotein (P-gp). Upon exposure to chemotherapeutics, drug-resistant cells increase release of ABCB1-carrying EVs, mediated by Rab8B.Furthermore, in drug-sensitive tumor cells, Rab5 downregulation accelerated the recycling of these exosomes to the plasma membrane. This process makes it easier for ABCB1 to move between cells, which allows normally susceptible cells to quickly develop resistance. This resistance is temporary, though, and mainly helps cells avoid the cytotoxic effects of chemotherapy medications (39).

Beyond drug efflux, cytochrome P450 (CYP) enzyme-mediated metabolism of chemotherapeutics has recently been identified as a key biological driver of treatment resistance in cancer. Chemotherapeutic medicines can be reduced in cytotoxicity by being converted into inactive or low-toxicity metabolites by the CYP enzyme system, which is essential for drug metabolism (40). These exosomes deliver enzymes to target cells via membrane fusion or internalization, further enhancing drug metabolism (41).

Chemoresistance in HCC is a complex multifactorial process, with the exosome-mediated non-coding RNA (ncRNA) regulatory network emerging as a key breakthrough in mechanistic studies in recent years. Exosomes remodel the tumor microenvironment and activate drug resistance-associated signaling pathways by delivering ncRNA molecules (e.g., miRNAs, lncRNAs, circRNAs) (42). To regulate post-transcriptional alterations in gene expression, exosomes generated from HCC cells deliver these non-coding RNAs to target cells via endocrine or paracrine pathways. Exosomes, for example, have the ability to move circDCAF8 from HCC cells that are resistant to regorafenib to those that are susceptible to it. Regorafenib-sensitive cells can develop a drug-resistant phenotype through this translocation process (43). Similarly, circRNA-SORE is translocated by exosomes to propagate sorafenib resistance in HCC cells (44). Further studies indicate that exosomes from sorafenib-resistant HCC cells are enriched with circUPF2, which enhances sorafenib resistance by promoting SLC7A11 expression and inhibiting ferroptosis.circUPF2 enhances sorafenib resistance by stabilizing the circUPF2-IGF2BP2-SLC7A11 ternary complex, which stabilizes SLC7A11 mRNA, promotes systemic Xc^- function, and reduces ferroptosis sensitivity, thereby inducing drug resistance (45). Exosomal circCCAR1, when taken up by CD8+ T cells, contributes to CD8+ T cell dysfunction by stabilizing PD-1 proteins, which in turn leads to resistance to PD-1 immunotherapy (46). Notably, exosomes can enhance cellular autophagy, thereby increasing tumor cell tolerance. For example, circTGFBR2, which is transported to HCC cells by exosomes, acts as a competitive endogenous RNA by binding miR-205-5p, which increases autophagy and ATG5 expression, making HCC cells more resistant to starvation (47) ( Figure 1 ).

Primary tumors induce adaptive alterations in the microenvironment of distant organs prior to metastasis, creating favorable conditions for tumor colonization and growth—a process referred to as pre-metastatic niche (PMN) formation (48). By modifying hypoxia, promoting angiogenesis, and reprogramming tumor metabolism, exosomes help establish the PMN, thereby facilitating metastasis by preparing secondary organs to support circulating tumor cells and promote their survival and proliferation (49).

Hypoxia is a key component of the microenvironment in solid tumors, particularly hepatocellular carcinoma (HCC). Both hypoxia and the consequent oxidative stress play important roles in the progression and treatment of HCC. Aside from increasing tumor cell invasiveness and stem-like qualities, hypoxic circumstances also drive cancer cells to create more cytokines, non-coding RNAs, and exosomes (50, 51). For instance, hypoxia-exposed HCC cells package miR-1290 into exosomes, which facilitates M2 macrophage polarization by downregulating Akt2 and upregulating PD-L1. This phenotypic shift in macrophages induces CD8+ T cell apoptosis and promotes epithelial-mesenchymal transition (EMT), thereby enhancing HCC cell migration (52). Likewise, miR-4508 expression is elevated in tumor-derived exosomes (TDEs) under hypoxic stress. This miRNA targets the 3′ untranslated region (3′UTR) of regulatory factor X1 (RFX1), leading to its downregulation. Reduced RFX1 levels activate fibroblasts and promote pre-metastatic niche (PMN) formation via the IL17A–p38 MAPK–NF-κB signaling cascade. These activated fibroblasts subsequently aid in the recruitment and expansion of myeloid-derived suppressor cells (MDSCs). Consequently, the lung PMN established via fibroblast activation facilitates metastatic spread of primary tumors to the lungs by establishing a favorable tumor colonization environment (53).

HCC frequently undergoes metabolic reprogramming, characterized by enhanced aerobic glycolysis—a phenomenon termed the Warburg effect.This metabolic pattern effectively generates energy and vital macromolecules for cell growth, thereby driving swift tumor expansion and proliferation. It was discovered that tumor-associated macrophage (TAM)-derived exosomes containing lncMMPA play important roles in macrophage polarization, HCC glycolytic pathway activation, and cell proliferation. In particular, TAM-derived lncMMPA promotes M2-type macrophage polarization. Elevated lncMMPA levels in HCC promote cell proliferation and aerobic glycolysis by stabilizing aldehyde dehydrogenase 1A3 (ALDH1A3) (54). In HCC, tumor exosomes carrying fatty acids—particularly palmitic acid—stimulate hepatic macrophages to secrete tumor necrosis factor (TNF), leading to a pro-inflammatory microenvironment that impairs fatty acid metabolism and oxidative phosphorylation, thereby promoting fatty liver progression (55). In the pre-metastatic niche, cancer cells secrete miR-122-rich vesicles to restrict glucose uptake by non-tumor cells (56). Interestingly, it has been shown that breast cancer (BC) cells inhibit pyruvate kinase (PKM) in pancreatic islet β-cells by secreting miR-122-rich extracellular vesicles, which in turn reduces insulin secretion and leads to dysregulation of glucose homeostasis. In addition to encouraging tumor growth, this mechanism raises the risk of type 2 diabetes in BC patients (57).

The advancement and dissemination of HCC are significantly contingent upon angiogenesis. Sustained multiplication of tumor cells is dependent on an appropriate supply of nutrients and oxygen, and the neovascular network formed by angiogenesis is vital for meeting this demand. Exosomes secreted by HCC cells is crucial to pathological angiogenesis by delivering various pro-angiogenic molecules (58). Golgi protein 73 (GP73), which is significantly upregulated in HCC, stimulates the production and secretion of vascular endothelial growth factor A (VEGFA) and promotes angiogenesis in the tumor microenvironment (59). According to clinical studies, individuals with HCC and high miR-3174 expression have a worse prognosis. Under hypoxia, miR-3174 is increasingly packaged into exosomes, facilitating angiogenesis and metastasis in HCC by suppressing the HIPK3/p53 and HIPK3/Fas signaling cascades (60). Similarly, exosome-derived long non-coding RNA LINC00161 targets and inhibits miR-590-3p, thereby relieving the inhibitory effect of miR-590-3p on its downstream target gene ROCK2 and activating the ROCK2 signaling pathway. Stimulation of this pathway significantly enhances HCC angiogenesis (61). Concurrently, circRNA - 100,338 is abundantly present in metastatic hepatocellular carcinoma (HCC) cells and their exosomes. Its upregulation or downregulation notably affects HCC cell invasiveness, endothelial cell proliferation, angiogenesis, and permeability, all contributing significantly to tumor metastasis (62). In conclusion, acquiring a thorough understanding of the mechanisms driving pre-metastatic niche (PMN) formation is crucial for developing effective therapies. This knowledge will enable us to target early metastatic stages and deepen our understanding of organ-specific metastasis (63).

Since Stephen Paget proposed his hypothesis in 1889, the organotropism of pre-metastatic niche (PMN)-mediated metastasis has remained one of the most puzzling mysteries in cancer biology (64). The underlying mechanisms of this enigma may involve several factors. Firstly, Tumor exosomes can recruit immunosuppressive cells. After upregulating pro-inflammatory factors, exosomes induce a local inflammatory microenvironment that prompts tumor cells to produce chemokines and cytokines (65). These factors, together with tumor-derived exosomes, recruit TAMs, TANs, Tregs, and MDSCs to distant sites, potentially suppressing anti-tumor immunity,which may facilitate tumor cell metastasis and colonization (66). Furthermore, factors secreted by primary tumor cells can activate fibroblast-like stromal cells at pre-metastatic sites, promoting increased synthesis of the extracellular matrix (ECM) component fibronectin (67). Interestingly, recent studies have unveiled the critical role of exosomal integrins in mediating the targeted metastasis of exosomes, providing a novel perspective for deciphering the precise targeting mechanisms of tumor metastasis. The studies confirms that exosomes secreted by lung, liver, brain-tropic tumor cells preferentially fuse with resident cells in target organs (such as lung fibroblasts and liver Kupffer cells), remodeling tumor metastatic pathways by establishing pre-metastatic niches. For instance, lung-tropic exosomes can redirect the metastatic trajectory of bone-tropic tumor cells. Proteomic analysis of exosomes reveals that integrins α6β4/α6β1 are specifically associated with lung metastasis, while integrin αvβ5 is linked to liver metastasis. Targeting these integrins can significantly inhibit exosome uptake and organ-specific metastasis. Mechanistic studies show that integrin-mediated exosome uptake activates Src phosphorylation and induces the expression of pro-inflammatory S100 genes. Clinical data further indicate that the expression profile of exosomal integrins can serve as a potential biomarker for predicting organ tropism of tumors (64, 68, 69) ( Figure 2 ).

Immune checkpoint molecules, acting as “regulatory hubs” for immune homeostasis, possess physiological inhibitory functions that originally evolved to prevent autoimmunity. However, they are systematically hijacked by tumor cells, becoming key pathways for immune evasion. These molecules maintain self-tolerance by suppressing immune cell activity, while tumor cells act. ivate this inhibitory network through upregulating ligands like PD-L1, driving T cells into anergic and exhausted states to ultimately evade immune surveillance (72, 73). Within the identified inhibitory immune checkpoints, programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) are prominent regulators (74). Preclinical studies have also identified additional immune checkpoints on circulating small extracellular vesicles (sEVs), including PD-1, CTLA-4, and CD80 (75). The abnormal regulation of PD-L1 in tumor cells remains largely unclear. In the study, researchers systematically explored the endosomal transport of cell-surface PD-L1 in tumor cells. Their findings reveal that plasma membrane PD-L1 undergoes constant internalization, followed by trafficking from early endosomes to multivesicular bodies/late endosomes, recycling endosomes, lysosomes, and/or extracellular vesicles (EVs) (76). And disrupting phosphatidylinositol-4-phosphate (PI(4)P) generation or exosome complex function blocks exosomal secretion of the key immune checkpoint protein programmed death-ligand 1 (PD-L1) in tumor cells, leading to its accumulation in lysosomes (77). By comparing the latest studies, it is revealed that, the regulation of PD-L1 by immune cell-derived small extracellular vesicles (I-sEVs) and triazine compound 6J1 exhibits both opposing and complementary biological effects. PD-1/CD80 carried by I-sEVs promotes the secretion of PD-L1+ exosomes by tumor cells and reduces cell-surface PD-L1 expression, simultaneously inhibiting antigen-presenting molecules and intercellular adhesion molecules to ultimately induce an immunosuppressive “cold tumor” phenotype (78). In contrast, 6J1 blocks PD-L1 endocytic recycling by activating Rab5, leading to its accumulation in endocytic vesicles, while activating Rab27 to promote PD-L1 exosomal secretion (76). These dual mechanisms synergistically reduce membrane PD-L1 density, not only enhancing tumor cells’ sensitivity to T-cell killing but also remodeling the immune microenvironment by increasing tumor-infiltrating cytotoxic T cells and chemokine secretion.

CD8 + T cells are crucial for tumor immunity and serve as primary antitumor defense cells. Tumor tissues can deplete CD8⁺ T lymphocytes through several mechanisms, decreasing their antitumor efficacy and allowing immunological escape (79). For instance, HCC cells release circCCAR1 in an hnRNPA2B1-dependent manner. The secreted circCCAR1 is then internalized by CD8+ T cells, inducing their dysfunction and consequently contributing to resistance to anti-PD-1 immunotherapy (46). Meanwhile, through encouraging the polarization of M2-type macrophages, miR-1290-containing exosomes released by HCC cells in a hypoxic environment triggers CD8⁺ T cell death (80). Exosomes also significantly influence CD4+ T lymphocyte differentiation. For instance, some exosomes promote the differentiation of naive CD4+ T cells into helper T cell (Th) subsets (e.g., Th1, Th2, Th17), while others induce the generation of regulatory T cells (Tregs). This differentiation mechanism is critical for regulating immune responses and maintaining immune homeostasis (81).

Furthermore, miR-500a-3p carried by HCC exosomes is involved in immune cell differentiation, HCC growth and invasion, and hepatic stellate cell (HSC) activation. Compared with the miR-500a-3p agonist-resistant group, the group treated with the miR-500a-3p agonist showed a significantly higher proportion of Tregs, highlighting the crucial role of miR-500a-3p in the suppressive microenvironment and T cell activity (82). Notably, aside from IL-2, extracellular vesicles (EVs) act as key intermediaries between CD4+ and CD8+ T cells. CD4+ T cell-derived EVs carry miR-25-3p, miR-155-5p, miR-215-5p, and miR-375, all of which activate CD8+ T cells. In a mouse melanoma model, these EVs have been shown to inhibit tumor progression by activating CD8+ T cells (70). Additionally, adipose-derived stem cells (ASCs) release exosomes that promote breast cancer progression. Zhu et al. investigated the effects of ASC-derived exosomes on gene and protein expression using western blotting, qRT-PCR, and RNA-seq. CD4+ T cells can internalize ASC-derived exosomes, which then promote their differentiation into Tregs (83).

Natural killer (NK) cells are integral to tumor immunosurveillance, as they directly eliminate mutant cells via death receptors and cytotoxic granules (84). However, in the tumor microenvironment (TME), NK cells often exhibit functional exhaustion, impairing their immunosurveillance capacity and enabling tumor cells to escape immune monitoring, thereby promoting tumor growth (85, 86). In HCC patients, HCC cells secrete circUHRF1 via exosomes, which inhibits natural killer (NK) cell function (87). As indicated in studies, NEAT1 in exosomes secreted by multiple myeloma cells suppresses NK cell activity and facilitates immune evasion by silencing PBX1 and recruiting EZH2 (88). In plasma exosomes from gastric cancer patients, elevated miR-552-5p expression is inversely correlated with the phenotypic distribution and receptor activation of NK cells. This miRNA reduces NK cell secretion of perforin, granzyme, and IFN-γ, and downregulates the expression of activation receptors (NKp30, NKp46, NKG2D), thereby impairing NK cell cytotoxicity.

Myeloid-derived suppressor cells (MDSCs) are heterogeneous myeloid cells originating from the bone marrow. Phenotypic differences in MDSCs from various cancer tissues and organs suggest that their differentiation and accumulation may be dependent on cancer cells (89). For instance, recent studies have revealed that exosomal PD-L1 promotes MDSC proliferation by activating the IL-6/STAT3 signaling pathway in vitro. In a CDX model, exosomal PD-L1 was shown to facilitate tumor progression and enhance MDSC expansion, highlighting its role in immunosuppressive microenvironment formation (90).

In cancer patients, particularly those in advanced stages, neutrophil accumulation in the peripheral blood is significantly increased. A high circulating neutrophil-to-lymphocyte ratio (NLR) is associated with poor prognosis in various cancers. Over the past decade, neutrophils’ role in cancer has gained considerable attention (91). In the TME, tumor-infiltrating neutrophils are capable of polarizing into anti-tumor N1 subsets that facilitate T cell-mediated tumor eradication or pro-tumor N2 subsets that serve as immunosuppressive cells (92). It has been confirmed that cancer cells induce neutrophil polarization via exosomal pathways, driving their conversion to pro-tumor phenotypes phenotypes.A research showed that exosomes derived from colorectal cancer (CRC) facilitate the circulation of PACRGL and promote neutrophil differentiation to the N2 phenotype via the miR-142-3p/miR-506-3p-TGF-β1 axis. Knockdown of CircPACRGL impairs this differentiation, which can be reversed by exosome complementation (93).

At the same time, macrophages play dual roles in cancer progression and metastasis. Pro-inflammatory M1 macrophages can phagocytose tumor cells and exert antitumor effects, whereas anti-inflammatory M2 macrophages—particularly tumor-associated macrophages (TAMs)—promote tumor growth and invasion by secreting cytokines and growth factors (94). Exosomes contain non-coding RNAs (ncRNAs), which are essential for a variety of physiological functions. Tumor-derived exosomal ncRNAs polarize macrophages toward the M2 phenotype by activating signaling pathways, regulating signal transduction, modulating transcriptional and post-transcriptional processes, and promoting the formation and progression of the tumor microenvironment (95). Upon uptake by human umbilical vein endothelial cells and HCC cells, M2 macrophage-derived exosomes stimulate angiogenesis, vascular permeability, and epithelial-mesenchymal transition (EMT). A feedback loop is established when HCC cells cocultured with M2 exosomes release increased levels of GM-CSF, VEGF, G-CSF, MCP-1, and IL-4, which recruit M2 macrophages (96). Jia et al. discovered that HCC-derived exosome circTMEM181 increases immunosuppression and resistance to anti-PD-1 reatment by upregulating CD39 expression. Targeting CD39 on macrophages can inhibit the ATP-adenosine pathway and reverse this resistance (97). Tumor-associated macrophages (TAMs) enhance the aggressiveness and biological behavior of HCC cells through multiple mechanisms. TAMs secrete exosomes containing various molecules that promote intercellular communication in HCC and modulate target cell functions. Non-coding RNAs also regulate TAM polarization in the tumor microenvironment (TME). These findings suggest that TAMs and their exosomes play a critical role in establishing the immunosuppressive microenvironment of HCC and could serve as potential therapeutic targets (98).

Tumor cells have been identified for their ability to release exosomes, which have a substantial impact on the tumor microenvironment (TME). However, it is also important to recognize that stromal cells within the TME are prolific exosome producers. These stromal cells, which include cancer-associated fibroblasts (CAFs), and endothelial cells, Mesenchymal Stem Cells (MSCs) produce exosomes containing a variety of bioactive chemicals that influence both tumor cell behavior and immune cell function, thereby contributing to the formation of an immunosuppressive microenvironment.

The majority of eukaryotic cells produce exosomes, a form of extracellular vesicle that is essential for intercellular communication. These vesicles carry numerous bioactive molecules, including proteins, DNA, mRNA, microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and circular RNAs (circRNAs). Exosomal cargo plays a major role in tumor growth, metastasis, and angiogenesis over the course of cancer, and these elements have potential as predictive biomarkers in cancer patients (120). Table 1 summarizes clinical trials focused on exosome-based detection of digestive system cancers.In particular, exosomal miRNAs have demonstrated diagnostic potential for HCC (121). For example, Wang et al. identified a panel of exosomal miRNAs—miR-122, miR-21, and miR-96—that could serve as reliable diagnostic biomarkers for HCC (122). Furthermore, reduced expression of exosomal miR-23a-3p has been found to be negatively correlated with overall survival in HCC patients, suggesting its potential for evaluating prognosis and treatment response (123). Although the clinical application of such RNA-based biomarkers for HCC screening remains to be fully validated, their potential is considerable. Exosomes have already been employed in clinical trials to detect malignancies within the digestive system, underscoring their diagnostic promise. However, data specific to exosome-based identification in HCC are currently limited. Continued research in this area is anticipated to yield significant breakthroughs, advancing early detection and precision therapy for HCC.

Notably, multiple studies have constructed and validated hepatocellular carcinoma prognostic models based on exosomal mRNA using multiple databases, identifying key genes that not only have diagnostic value but are also closely associated with immunotherapy and the tumor microenvironment (124). For example, one study addressed the challenge of prognostic evaluation in HCC by exploring the value of a risk prognostic model based on blood exosomal mRNA. From the TCGA and exoRBase 2.0 databases, 44 prognosis-related genes were screened, and 6 exosomal risk genes including CLEC3B were selected through analysis to construct the model. Validation showed that the model’s risk score was a robust independent prognostic factor, and the nomogram model incorporating pathological stage yielded the best clinical benefits. This study is the first to confirm that these 6 genes exist in blood exosomes of HCC patients, enabling liquid biopsy to avoid puncture. These genes originate from multiple cell types, potentially serving as diagnostic markers, and the high-risk group may benefit from immunotherapy (125) Additionally, A 4-gene prognostic signature (DYNC1H1, PRKDC, CCDC88A, and ADAMTS5) related to exosomal mRNA was constructed and validated. A prognostic nomogram based on this signature exhibits prognostic ability for HCC. The genes in the signature are involved in the extracellular matrix, ECM-receptor interaction, and PI3K-Akt signaling pathway. Their expression is positively correlated with immune cell infiltration in the TME (126).

Researchers are interested in creating engineered exosome nanocarriers because exosomes are important for cellular communication and, as naturally occurring nanocarriers, they are not only low-immunogenic and biocompatible but also have a lot of potential for pharmacotherapy and drug targeting (127). Table 2 presents a comparison of Natural Exosomes, Engineered Exosomes, and Traditional Carriers.

On the one hand, exosome targeting can be enhanced by surface modification (128, 129), thus enabling the precise treatment of specific tumor sites or tissues. For example, by enhancing targeting, it can be used to treat brain, breast, lung, liver, colon tumors, and heart diseases (129). By modifying the exosome surface with an EGFR antibody, Ohno et al. and Kooijmans et al. found that exosomes could be targeted to EGFR-expressing tumor cells (130). Tamura et al. used electrostatic interactions to modify exosomes with cationic branched-chain amyloid polysaccharides, which increased their uptake in HepG2 cells considerably. Unlike unmodified exosomes, modified exosomes enter cells via the desialylated glycoprotein receptor, which is exclusively expressed by hepatocytes. When modified exosomes were intravenously administered into animals with concurrent sabinoglobulin A-induced liver injury, they accumulated in the liver tissues and dramatically increased the anti-inflammatory effects (131). The use of exosomes as drug carriers has attracted much attention. Owing to their non-immunogenicity, low toxicity, and biocompatibility, they are considered ideal drug delivery vehicles (132). For example, Mukhopadhya et al. demonstrated, through the application of high-performance liquid chromatography (HPLC), that the electroporation method successfully encapsulated 90.1 ± 12 μg (36% of the available adriamycin) and 68.0 ± 10 μg (27%) of the drug into MSCs and milk exosomes, respectively, in a comparative study of drug loading into milk and MSC exosomes. Using imaging flow cytometry, they confirmed that this method did not compromise the integrity of exosomal surface proteins, whereas the sonication method significantly reduced the number of CD9+/CD63+ exosomes (p<0.0001), thus validating the efficacy and safety of exosomes as drug carriers (133). BM-MSC exosomes have also made significant progress in the treatment of HCC as drug carriers. BM-MSC-derived exosomes delivering the nucleic acid drug siGRP78 enhanced the sensitivity of HCC cells to sorafenib, thereby improving the drug effect (134). Extracellular vesicle (EV) proteomics analysis can help predict therapy response. In a rare clinical trial, 25 patients with advanced HCC were treated with selective internal radiation therapy (SIRT) in combination with sorafenib, while 20 patients received sorafenib alone. The results revealed that patients with greater EV-GPX3/ACTR3 levels and lower EV-ARHGAP1B levels had superior efficacy when treated with SIRT in conjunction with sorafenib (AUC = 1) (135) ( Figure 4 ).