A study reported that the expression levels of circular RNA (circ)-kinesin family member 4A (KIF4A) in TNBC brain metastases and cell lines are markedly upregulated compared with those in primary TNBC tissue and normal human mammary epithelial cells, respectively. Although no significant difference in miR-637 expression was observed between primary breast cancer and brain metastasis (15), miR-637 expression was found to be notably reduced in brain metastasis. This may be attributed to circKIF4A directly binding to miR-637, thereby acting as a competing endogenous RNA by sponging miR-637, regulating the miR-637-mediated regulation of STAT3 expression and promoting the metastasis of breast cancer to the brain (15). Whilst the mechanism by which circKIF4A reaches the brain remains unclear, this finding highlights a potential regulatory axis involved in breast cancer metastasis and warrants further investigation.

One key mechanism underlying breast cancer brain metastasis is penetration of the blood-brain barrier (BBB). Extracellular vesicles (EVs), particularly exosomes, have been reported to facilitate this process. Exosomal miR-105 preferentially accumulates in brain microvessels and downregulates the expression of tight junction protein zonula occludens-1 (ZO-1); this is supported by clinical evidence demonstrating that elevated exosomal miR-105 levels are negatively associated with ZO-1 (10). The accumulation of exosomal miR-105 increases vascular permeability, which creates conditions favorable for tumor cell extravasation (10). In addition, miR-105 inhibits suppressor of cytokine signaling 1 (SOCS1), relieving its suppression of the JAK-STAT pathway and promoting STAT3 phosphorylation (16). miR-105 also activates the NF-κB signaling pathway, upregulates vascular cell adhesion molecule 1 (VCAM1) expression and increases the exposure of endothelial cell adhesion molecules, further promoting tumor cell adhesion and extravasation (17–20).

Breast cancer cells secrete exosomes and EVs into the circulatory system, which are then taken up by brain endothelial cells via transcytosis, crossing the BBB to enter the brain microenvironment (21). Elevated VCAM1 expression recruits M2-type macrophages, which secrete factors such as IL-6 and TNF-α, thereby remodeling the immune microenvironment (18,19,22), and enabling tumor cells to traverse the BBB and enter brain tissue. Research has reported that breast cancer cells secrete exosomes carrying miR-19a, which recognize receptors on brain microvascular endothelial cells via surface integrin αvβ3, crossing the BBB via transcytosis and preferentially enriching around astrocytes (23). In astrocytes, miR-19a directly inhibits the expression of the tumor suppressor genes PTEN and SOCS1 (23). PTEN loss activates the PI3K-AKT-mTOR pathway, thereby promoting tumor cell survival, whilst SOCS1 inhibition leads to sustained STAT3 phosphorylation. Phosphorylated STAT3 induces the expression of IL-6 and TGF-β, which maintain tumor stem cell properties and induce the construction of a fibrotic matrix that facilitates tumor adhesion and colonization (24–27).

After colonization, breast cancer cells, particularly those from TNBC, secrete exosomes containing miR-122 in brain tissue (28). miR-122 directly binds to the 3′ untranslated region (3′-UTR) of pyruvate kinase M2 (PKM2) mRNA, inhibiting its translation and leading to a ~70% reduction in PKM2 activity, which disrupts the final step of glycolysis (29). This causes glycolytic intermediates such as phosphoenolpyruvate (PEP) and 3-phosphoglycerate (3-PG) to accumulate and divert into the pentose phosphate pathway (PPP) and serine synthesis pathway, promoting the production of NADPH and nucleotide precursors that support tumor proliferation (30,31). Consequently, this metabolic reprogramming establishes an energy-rich, antioxidant-enriched and immunosuppressive microenvironment, supporting sustained tumor cell colonization (29,32,33).

Bone metastasis is one of the most challenging complications of breast cancer, and is primarily driven by osteoclast activation (34), osteoblast inhibition (35,36) and the osteolytic cycle (37). These mechanisms are central to the development of bone metastasis and are closely associated with miRNAs (38).

Osteoprotegerin (OPG), receptor activator of nuclear factor κB (RANK) and RANK ligand (RANKL) constitute the key signaling pathway regulating bone homeostasis (39). Activation of this pathway by several factors, including TNF-α, IL-1/11, RANKL, macrophage colony-stimulating factor and endocrine and metabolic-related factors leads to bone loss due to the disruption of bone homeostasis, resulting in an imbalance between bone formation and resorption (40). RANKL is a key regulator of osteoclast differentiation and proliferation. In the RANKL/RANK/OPG axis, RANKL interacts with its receptor RANK on osteoclast precursors, ultimately leading to their maturation. By contrast, OPG acts as a decoy receptor for RANKL, thereby inhibiting the RANKL-RANK interaction and mediating bone remodeling (41–43).

RUNX family transcription factor 2 (RUNX2), a key transcription factor in osteoblast differentiation, functions as a transcriptional activator of OPG. Reduced RUNX2 expression leads to decreased OPG levels, thereby accelerating osteolysis (44). The overexpression of miR-218-5p directly suppresses RUNX2; moreover, evidence indicates that miR-218-5p also directly binds to the OPG 3′-UTR, further reducing OPG expression (31). In this context, TNBC cells have been shown to deliver high levels of miR-218 via exosomes (45). Elevated miR-218 inhibits the bone remodeling function of OPG and consequently activates the RANKL/RANK pathway, thereby promoting osteolysis.

Another miRNA, miR-34, serves a key role in osteoblast inhibition during breast cancer bone metastasis. Osterix (Osx) is a core transcription factor for osteoblast differentiation and regulates the synthesis of bone matrix proteins, including type I collagen and osteocalcin. Osx deficiency has been reported to be associated with a ~70% reduction in mineralized nodule formation and a marked impairment in bone repair capacity (46,47). It has been reported that miR-34c directly binds to the 3′-UTR of Osx mRNA, mediating its degradation via the RNA-induced silencing complex, thereby inhibiting Osx protein expression (48–50). In addition to this direct effect, miR-34c also targets large tumor suppressor kinase 1, inhibiting its phosphorylation and inducing the nuclear translocation of Yes-associated protein and transcriptional coactivator with PDZ-binding motif, which further suppresses osteoblast differentiation (51). In addition, Osx deficiency indirectly attenuates Wnt pathway activity by reducing β-catenin nuclear translocation efficiency, ultimately diminishing osteogenesis (52). Therefore, through both direct binding and indirect mechanisms, miR-34c downregulates Osx, impairs bone matrix repair and exacerbates the cancer cell-induced destruction of bone.

Beyond osteoclast activation and osteoblast inhibition, miRNAs play a major role in the osteolytic cycle. Exosomes secreted by breast cancer cells carry miR-21, which targets osteoclast precursors and mesenchymal stem cells in the bone microenvironment via surface integrins. Following uptake, the exosomes release large amounts of miR-21, which directly suppress the expression of programmed cell death 4 (PDCD4), a tumor suppressor gene, in osteoclasts. This suppression activates nuclear factor of activated T cells 1 (NFATc1) signaling, thereby promoting osteoclast differentiation (37). As osteoclast activity intensifies and bone matrix degradation progresses, osteoclasts release TGF-β, forming concentration gradients in the tumor microenvironment that attract breast cancer cells toward the resorption sites (53). This resulting accumulation of tumor cells and TGF-β facilitates the binding of TGF-β to the TGF-β receptor type II on tumor cell surfaces, activating the SMAD2/3 signaling pathway. This activation induces tumor cells to secrete insulin-like growth factor-1 (IGF-1) and platelet-derived growth factor (PDGF), which promote tumor proliferation and survival, as well as tumor angiogenesis and fibroblast activation (54). Increased tumor burden leads to increased miR-21 secretion, which accelerates the osteolytic process by promoting osteoclast-mediated bone resorption and leading to further TGF-β release from the bone matrix. Elevated TGF-β, in turn, upregulates miR-21 transcription via SMAD4 binding to the miR-21 promoter, thereby forming a positive feedback loop. This reciprocal reinforcement leads to persistently high TGF-β concentrations in the bone microenvironment (55). Concurrently, TGF-β induces miR-19a expression, which targets PTEN in osteoblasts, inhibiting bone matrix repair and exacerbating osteolytic destruction (56). In summary, exosomal miR-21 drives the malignant osteolytic cycle of breast cancer bone metastasis through activation of the PDCD4/NFATc1 axis in osteoclasts, leading to bone matrix degradation, the release TGF-β, induction of IGF-1 and PDGF secretion, and subsequent tumor proliferation’.

The liver is a common and lethal site of metastasis in TNBC, involving a complex regulatory network in which EVs, particularly exosomes, and miRNAs together modulate metabolism, immune responses and microenvironmental homeostasis (31). miR-122 serves as a central metabolic regulator in this process. Tumor-derived exosomes deliver miR-122 to non-tumor cells in distant organs such as the liver, thereby executing a metabolic reprogramming strategy by directly targeting and inhibiting the translation of PKM2 (29). PKM2 is a key rate-limiting enzyme in glycolysis. Its inhibition reduces hepatic glycolytic capacity, forcing increased glucose availability for tumor cells, while also leading to the accumulation of glycolytic intermediates, including PEP and 3-PG. These intermediates are subsequently diverted into the PPP and the serine synthesis pathway, generating NADPH and nucleotide precursors that support metastatic growth, redox balance and biosynthetic demands (30,31). This metabolic rewiring enhances the Warburg effect, characterized by aerobic glycolysis within tumor cells, supplying energy and biosynthetic precursors required for metastatic growth (29,57).

Circ-phosphoglycerate dehydrogenase (PHGDH) promotes aerobic glycolysis by binding to miR-122-5p, thereby relieving the miR-122-5p-mediated repression of PKM2. Conversely, silencing circPHGDH restores the miR-122-5p-mediated suppression of PKM2, resulting in reduced glycolytic activity and invasive capacity in tumor cells (58).

In addition to its local metabolic functions, miR-122 supports metastasis by inducing systemic cachexia. miR-122 packaged within EVs targets O-GlcNAc transferase (OGT) in skeletal muscle, leading to a reduction in the OGT-mediated O-GlcNAcylation of downstream substrates, (59), particularly that of sarcoplasmic reticulum ryanodine receptor 1, thereby increasing its protein abundance and elevating intracellular calcium ion (Ca²⁺) levels. The Ca²⁺ overload activates calpain, which degrades myofibrillar structural proteins, including desmin, and ultimately results in muscle wasting (59). This process releases nutrients such as amino acids into the circulatory system, indirectly providing metabolic substrates for liver metastasis growth.

miR-122 also directly disrupts the intrinsic microenvironment of the liver. The absence of miR-22 in hepatocytes upregulates the bile acid synthase hydroxy-δ5-steroid dehydrogenase, 3 β- and steroid δ-isomerase 7, leading to bile acid accumulation, while simultaneously downregulating hepatic nuclear factor 4α. These alterations reshape the hepatic microenvironment in a manner that promotes breast cancer metastasis (60).

The establishment of an immunosuppressive microenvironment represents another key pillar of liver metastasis, primarily driven by exosomal miR-21. Following the uptake of miR-21-carrying breast cancer exosomes by hepatic Kupffer cells, miR-21 suppresses NF-κB activation downstream of the Toll-like receptor 4 signaling pathway. This suppression reduces the release of pro-inflammatory factors, including TNF-α, thereby weakening the antitumor immune response (61). Concurrently, miR-21 induces the polarization of macrophages/Kupffer cells toward an immunosuppressive M2 phenotype, promoting the secretion of immunosuppressive cytokines such as IL-10 (62,63). In a metastatic breast cancer mouse model, these changes were associated with a marked increase in the proportion of myeloid-derived suppressor cells within the liver, a ~60% reduction in T-cell proliferation capacity, and a substantial impairment of the antitumor immunity of the liver microenvironment (61,62). In addition to immunomodulation, tumor-derived EVs and other extracellular particles can induce hepatic metabolic dysfunction, which also facilitates metastasis (62).

The development of lung metastasis is driven by a sequential and collaborative miRNA network that orchestrates the following three phases: PMN formation, immune evasion and metastatic colonization.

Circulating exosomal miR-105 targets the tight junction protein ZO-1 in pulmonary vascular endothelial cells, compromising endothelial integrity and increasing vascular permeability, thereby facilitating tumor cell extravasation (10). Concurrently, exosomal miR-21 and miR-10b are internalized by resident pulmonary fibroblasts. miR-10b inhibits the translation of homeobox D10 (HOXD10), a transcriptional repressor of pro-metastatic genes, which leads to the derepression of downstream effectors such as RhoC and matrix metalloproteinases (MMPs) that enhance invasiveness (64,65). miR-21 targets tumor suppressor genes, including PTEN and PDCD4, resulting in the activation of key signaling pathways, including the PI3K/Akt and NF-κB pathways. Through these pathways, miR-21 promotes the activation of fibroblasts and their transformation into cancer-associated fibroblasts (CAFs) (66). Activation of the TGF-β/SMAD pathway further sustains this phenotypic switch and drives extracellular matrix remodeling. Activated CAFs secrete substantial amounts of IL-6, TGF-β and extracellular matrix components, remodeling the pulmonary interstitium into a pro-inflammatory and fibrotic PMN (66). Additionally, miR-24 contributes to the establishment of a pro-angiogenic and immunosuppressive microenvironment by downregulating the methyltransferase adenosine phosphomethyladenosine gene in pulmonary stromal cells (67).

Tumor-derived exosomes reaching the alveoli deliver miR-122 to alveolar macrophages (29). miR-122 inhibits glycolysis in these macrophages by suppressing PKM2, inducing a metabolic shift that promotes their polarization toward an immunosuppressive M2 phenotype, thereby weakening local immune surveillance (29). Conversely, tumor-associated macrophages (TAMs) secrete exosomes carrying miR-223-3p, which are internalized by breast cancer cells. miR-223-3p suppresses the chromatin regulator chromobox 5, leading to epigenetic derepression of pro-metastatic genes such as MMPs, consequently increasing tumor cell invasiveness (66).

The function of the miR-200 family exhibits spatiotemporal context dependence during metastasis (68). In the early stages of dissemination, high levels of miR-200 within tumor cells help maintain E-cadherin expression and an epithelial phenotype, potentially inhibiting initial detachment from the primary tumor (68). However, during later stages of metastatic colonization, miR-200 actively remodels the pulmonary microenvironment to support clonal expansion. A key mechanism involves the inhibition of the vesicular transport protein Sec23a, which suppresses the secretion of tumor suppressor factors, including insulin-like growth factor-binding protein 4 (IGFBP4) and tubulointerstitial nephritis antigen-like 1 (68,69). This effect complements the contact-induced upregulation of miR-199a-3p by mesenchymal stem cells, which enhances tumor stemness by suppressing the transcription factor forkhead box P2, thereby boosting metastatic potential (70). Concurrently, miR-9 induces pulmonary fibroblasts to release vascular endothelial growth factor, which promotes angiogenesis in metastatic foci (71).

Several miRNAs, including miR-10b, miR-21, the miR-200 family and the miR-221/222 cluster, are recurrently involved in multiple metastatic sites. However, their mechanistic contributions exhibit pronounced organ-specific adaptations (Table I).

miR-10b is a quintessential pro-metastatic miRNA. Its core mechanism across contexts involves inhibition of HOXD10, leading to the derepression of pro-invasive genes such as RhoC and MMPs (64,72). However, its downstream effects vary by organ. In liver metastasis, miR-10b primarily promotes intrahepatic colonization via activation of the PTEN/Akt axis (73,74), whereas in lung metastasis it contributes to CAF activation and PMN formation (66). By contrast, its roles in brain and bone metastasis remain incompletely characterized.

miR-21 functions as a central amplifier of metastatic progression. Its common mechanism is the targeting of tumor suppressors such as PDCD4 and PTEN (37,55). In bone metastasis, it drives the osteolytic cycle via the PDCD4/NFATc124 axis, whereas in liver metastasis it polarizes Kupffer cells to an M2 phenotype via the suppression of NF-κB signaling, which promotes immunosuppression (61,62). Additionally, in the lungs, miR-21 facilitates fibroblast activation and CAF formation (62). Although high levels of miR-21 have been associated with brain metastasis, the precise stromal targets are yet to be elucidated (75).

The miR-200 family epitomizes functional pleiotropy and context-dependency. While generally associated with maintenance of the epithelial phenotype, miR-200 family members display a temporal shift during metastasis. miR-200 supports lung and liver colonization by inhibiting Sec23a-mediated secretion of tumor suppressors, such as IGFBP4, as well as by regulating mesenchymal-epithelial transition (68,76). In the specific organ microenvironment of breast cancer brain metastasis, certain members of the miR-200 family (especially miR-141)- or an miRNA cluster composed of the miR-200 family and miR-29-actually promote brain metastasis by directly inhibiting the expression of the ADAM12 protein (77,78). Conversely, in bone metastasis, the downregulation of miR-429, another miR-200 family member, is associated with metastatic progression (79).

Finally, the miR-221/222 cluster promotes metastasis predominantly by targeting PTEN, activating the Akt pathway and downregulating E-cadherin (73,80). Strong associations have been reported with liver metastasis, where miR-221/222 enrich cancer stem cell (CSC) populations via PTEN/Akt signaling, and with lung metastasis, where they drive EMT by targeting the transcriptional repressor GATA binding 1 and the tumor suppressor PTEN (73,81). In brain metastasis, miR-221/222 compromises the BBB via E-cadherin loss (80). However, evidence supporting a role in bone metastasis is lacking.

In brain metastasis, miR-105 is characteristically expressed and secreted by metastatic breast cancer cells. Within the endothelial monolayer, tumor-secreted exosomes enriched in miR-105 accumulate, where miR-105 disrupts tight junctions and the integrity of natural metastatic barriers. The primary underlying mechanism is that miR-105 targets the endothelial tight junction protein ZO-1, thereby disrupting BBB permeability and facilitating tumor cell extravasation (10,82).

Another miRNA associated with brain metastasis is miR-19a, which activates the STAT3 signaling pathway in astrocytes, inducing the secretion of IL-6 and TGF-β release and forming a tumor survival-promoting microenvironment. Simultaneously, it suppresses glutamate transporter solute carrier family 1 member 2 in astrocytes, hijacking neural metabolism to promote tumor growth (82).

In bone metastasis, miR-218-5p inhibits OPG, thereby activating the RANKL/RANK/NFATc1 signaling pathway and enhancing osteoclast differentiation, leading to osteolytic destruction (31). Similarly, miR-21 targets the tumor suppressor PDCD4, leading to NFATc1 activation and increased osteoclast activation. In parallel, miR-21 induces the release of TGF-β from the bone matrix, which stimulates tumor cells to secrete IGF-1 and PDGF, thereby establishing a cycle from osteolysis to growth factor release and tumor proliferation (83,84).

In liver metastasis, miR-122-5p has been shown to regulate the mobility of breast cancer cells. Specifically, miR-122-5p is abundantly present in hepatocyte-derived exosomes, and is capable of suppressing syndecan-1 expression and increasing the invasive ability and survival of breast cancer cells. These findings indicate that the liver metastasis of breast cancer is highly likely to be associated with miR-122-5p (85).

miR-598-5p inhibits breast cancer growth and lung metastasis by targeting phosphatidic acid phosphatase type 2 domain-containing protein 1A (86). Similarly, miR-134 inhibits breast cancer lung metastasis by suppressing SLUG and the EMT markers E-cadherin and N-cadherin (87). By contrast, Ras-related protein Rab1A promotes lung metastasis by facilitating the sorting of the tumor-suppressive miR-200c into exosomes, thereby reducing the inhibitory effect of miR-200c within tumor cells. The exosomal miR-200c derived from metastatic cells suppresses the immune response of F4/80⁺ macrophages, thereby contributing to immune evasion. Notably, the administration of anti-Rab1A antibodies reduced the transport of miR-200c into exosomes and inhibited the metastasis of breast cancer to the lung (88). In addition, TAM-derived EVs have been reported to shuttle miR-660, which promotes the lung metastasis of breast cancer via activation of the Kelch like family member 21-mediated IKKβ/NF-κB p65 axis (89).

miRNAs serve important roles in the promotion of tumor growth and metastasis. Therefore, the identification of inhibitors of pro-metastatic miRNAs has become a major topic in the treatment of breast cancer.

Breast cancer metastasis suppressor 1 (BRMS1) suppresses metastasis in multiple tumor types without blocking tumorigenesis. Mechanistically, BRMS1 forms complexes with the transcriptional corepressor SIN3, histone deacetylases (HDACs) and selected transcription factors that alter metastasis-associated gene expression. Notably, BRMS1 upregulates the expression of several metastasis-suppressing miRNAs, including miR-146a, miR-146b and miR-335. Collectively, these findings indicate that BRMS1 coordinates the regulation of multiple metastasis-related miRNAs, potentially through the recruitment of BRMS1-containing SIN3:HDAC complexes to miRNA promoters, although the precise targets are yet to be identified (90).

Furthermore, miR-155 is upregulated in breast cancer. Research using soft agar colony formation assays and tumor xenograft models has demonstrated that the inhibition of miR-155 notably reduces cancer cell proliferation in vitro and in vivo, indicating that miR-155 may be a potential therapeutic target for breast cancer (91).

The tumor-suppressive let-7 family of miRNAs has been identified to induce apoptosis, inhibit proliferation and suppress the self-renewal capacity of CSCs. In a study assessing the inhibitory effect of let-7 miRNAs on the self-renewal capacity of TNBC CSCs, radiotherapy was found to suppress TNBC stem cell self-renewal by inhibiting cyclin D1 expression and Akt1 phosphorylation. Notably, let-7d enhanced the radiation-induced tumor suppression and synergized with radiotherapy to further inhibit CSC renewal. Western blotting, immunofluorescence and luciferase reporter assays suggested that reduced cyclin D1/Akt1/Wnt1 signaling activity contributed to the observed let-7-induced radiosensitization. Let-7 was shown to directly inhibit cyclin D1 expression, leading to hypophosphorylation of Akt1, and the suppression of mammosphere formation. Furthermore, let-7d-induced Akt1 inhibition exhibited tumor-suppressive effects comparable to those obtained with Akt inhibitors (92).

miR-708-3p has been reported to suppress EMT in breast cancer cells by directly targeting EMT activators, including zinc finger E-box binding homeobox 1, cadherin 2 (N-cadherin) and vimentin, thereby functioning as a cancer-suppressing miRNA in breast cancer (93). miR-381-3p expression is markedly downregulated in breast cancer tissues and cell lines. Functional studies have shown that the overexpression of miR-381-3p inhibits breast cancer proliferation and invasion in MDA-MB-231 and SKBR3 cells, whereas its knockdown promotes these behaviors. Mechanistically, miR-381-3p suppresses EMT by targeting Sox4 and Twist1 to regulate TGF-β signaling and inhibit breast cancer progression (94). Collectively, these findings support the upregulation of tumor-suppressive miRNAs as another potential therapeutic approach for the suppression of breast cancer progression.

The BBB inherently limits the entry of therapeutic drugs into the brain. Exosomes, a type of membrane-bound secreted lipid vesicle, are able to penetrate the BBB and may be used to deliver anticancer drugs to brain tumors at therapeutic levels. The capacity of exosomes to traverse the BBB is attributed primarily to transcytosis, an active transport mechanism involving endocytic uptake by brain endothelial cells, intracellular vesicular trafficking and exocytosis on the abluminal side (95–97). This process occurs predominantly via receptor-mediated transcytosis (RMT), wherein surface ligands on exosomes engage specific receptors on cerebral endothelial cells triggering internalization (97). Inflammation and the presence of metastatic tumor cells have been shown to enhance exosome trafficking across the barrier, although the precise molecular regulators remain under investigation (98). Engineered exosomes have been explored to exploit these pathways for targeted central nervous system (CNS) delivery. Engineering chimeric antigen receptor (CAR)-natural killer cell-derived exosome disguised nano-bombs for enhanced HER2-positive breast cancer brain metastasis therapy (99). These exosomes were dually modified: Surface expression of anti-HER2 single-chain variable fragment from CAR construct for tumor cell recognition, and conjugation with T7 peptide (HAIYPRH), which binds the transferrin receptor on brain endothelial cells. This design enhanced BBB penetration via RMT and achieved selective targeting of HER2-positive breast cancer brain metastases in orthotopic mouse models, significantly extending survival (99). Zhao et al (100) engineered exosomes from human adipose-derived mesenchymal stem cells (hAMSCs) modified to express anti-CD19 antibodies on their surface. These anti-CD19-Exos were loaded with methotrexate (MTX) and demonstrated enhanced BBB permeability in an in vitro model comprising hCMEC/D3 endothelial cells and astrocytes. In an intracranial CNS lymphoma model, systemically administered anti-CD19-Exo-MTX achieved sustained drug levels in cerebrospinal fluid, reduced disease burden and prolonged survival compared to free MTX or unmodified Exo-MTX. Mechanistic analysis indicated that exosome interaction with cerebrovascular endothelial cells and astrocytes facilitated endocytosis and subsequently facilitated the transportation of MTX across the barrier (100).

Although exosomes show promise for tumor-targeted therapy, several challenges remain. For example, exosome isolation and purification methods lack standardization, and variability in cargo composition, size and exosome source complicates their use in tumor therapy. In addition, achieving a high drug-loading efficiency in exosomes is challenging, and the scaling up of exosome production is a core bottleneck for this therapeutic approach (101).

Research indicates that therapeutic miR-10b inhibition in breast cancer treatment requires strict dosage control. A preclinical study revealed that a ~25-mg/kg dose (the drug is miR-10b antagomir, administered systemically, specifically via intravenous injection) is associated with elevated levels of the liver toxicity markers aspartate aminotransferase and alanine aminotransferase, whilst antitumor activity requires relatively high doses. This narrow therapeutic window is a major obstacle to the clinical translation of miR-10b-targeted therapies in breast cancer (102,103). These toxic effects may be attributable to the hepatic accumulation of oligonucleotides and potential immune activation (104).