SA, which involves the development of new capillaries from pre-existing blood vessels, has long been considered the predominant process underlying tumour vascularization (9). This intricate process is subject to multiple regulatory mechanisms governing vascular remodelling and maturation. These mechanisms encompass crucial signalling pathways, including: (1) The fibroblast growth factor (FGF) and its receptor (FGFR)signalling pathway: FGF and FGFR play pivotal roles in orchestrating angiogenesis by stimulating the growth and development of blood vessels. (2) The delta-like ligand 4 (DLL4)/Notch signalling pathway: DLL4 interacts with the Notch receptor, influencing cell fate decisions during angiogenesis and promoting vessel sprouting and branching. (3) The vascular endothelial growth factor (VEGF) and its receptor (VEGFR) signalling pathway: VEGF and VEGFR constitute a fundamental axis in angiogenesis, facilitating the formation of new blood vessels and promoting tumour angiogenesis. (4) The platelet-derived growth factor (PDGF) and its receptor (PDGFR) signalling pathway: PDGF contributes to the growth and maturation of blood vessels, playing a role in the tumour’s vascularization process (10–12). These signalling pathways are interconnected and interact to enhance the proliferation of vascular epidermal cells and induce tumour angiogenesis, creating favourable circumstances for malignant tumour development, invasion, and metastasis.

ECs demonstrate the ability to differentiate into two distinct phenotypes within the context of angiogenesis: tip cells, characterized by their capacity to migrate in response to a concentration gradient of the signalling molecule VEGF, and stalk cells, which are proficient in proliferative activities (13). In the intricate process of angiogenesis, the tip cells play a pivotal role by navigating toward the source of VEGF, guiding the extension of new vascular branches. These branches subsequently fuse with other neighbouring branches, culminating in the formation of a continuous vascular channel. This nascent vascular channel serves as the foundation for the subsequent development of a sophisticated network of blood vessels, encompassing capillaries, arterioles, and venous vessels. This orchestrated process significantly amplifies the extent of tumour angiogenesis (14), contributing to the increased supply of oxygen and nutrients to the tumour microenvironment, ultimately facilitating the progression and sustenance of malignant tumours.

Upon reaching a certain stage of growth, solid tumours are exposed to a hypoxic environment, which is considered a hallmark feature of tumour growth. Solid tumours generate a significant amount of Hypoxia-inducible factor 1 (HIF-1) under hypoxic conditions, and HIF-1 plays a crucial role in tumour progression, metastasis, and adverse prognosis. Research indicates that HIF-1 can activate the transcription of hypoxia-inducible genes by targeting genes’ promoters with a common DNA sequence known as hypoxia response elements (HREs). When induced, these HREs are activated, leading to the excessive secretion of various factors that promote angiogenesis (15). Essentially, the upregulation of HIF-1 in the hypoxia response plays a critical role in orchestrating molecular events that promote new blood vessel formation, providing essential oxygen and nutrients for the survival and development of tumours. Under hypoxic conditions, HIF-1 upregulates growth factors such as VEGF and FGF, promoting endothelial cell vascular permeability, endothelial cell growth, and proliferation. Furthermore, it can activate other growth factors, including PDGF, TGF, and the transcription of angiopoietins (Ang), facilitating the release of angiogenic factors by cancer cells.

VEGF is a proangiogenic factor that specifically targets vascular endothelial cells and is involved in a variety of events, such as tumour development, invasion, and angiogenesis. Members of the VEGF family include VEGF-A, VEGF-B, VEGF-C, VEGF-D, and placental growth factor (PIGF) (16). VEGF-A activates VEGFR-2, which leads to the upregulation of VEGFR-3 and Dll4 in tip cells and activation of the Notch pathway in neighbouring stalk cells. This process results in the downregulation of VEGFR-2 and VEGFR-3 and upregulation of VEGFR-1, maintaining the balance between tip and stalk cells ( Figure 2 ). The downstream pathways of the p38-mitogen-activated protein kinase (p38/MAPK) and AKT-phosphatidylinositol-3 kinases (PI3K/AKT) are activated upon VEGF-A binding to VEGFR-2 on the surface of ECs (17), causing ECs to proliferate and migrate, producing a large number of proangiogenic proteases that increase the permeability of the vessel barrier.

Once ECs have sprouted and aggregated to form a vascular prototype, they release factors to attract pericytes. Notably, heparin-binding EGF-like growth factor stimulates pericyte proliferation, while transforming growth factor beta 1/2 (TGFβ1/2) initiates pericyte recruitment (18). Additionally, by attracting pericytes and tumour-associated fibroblasts, PDGF is essential in the process of tumour angiogenesis. The PDGF family includes PDGF-A, PDGF-B, PDGF-C, and PDGF-D (19). PDGF binds PDGFR and promotes receptor dimerization, thereby initiating signalling, including the Ras-mitogen-activated protein kinase (RAS/MAPK) and PI3K pathways (20). PDGF-AB stabilizes the newly formed tumour vascular system by recruiting pericytes (21). PDGF-BB regulates VEGFR2 signalling and endothelial proliferation to prevent vascular abnormalities due to high levels of VEGF (22).

Ang induces tumour-endothelial cell interactions to regulate tumour angiogenesis. Ang-1, Ang-2, Ang-3, and Ang-4 are members of the Ang family, while Tie1 and Tie2 are their corresponding tyrosine kinase receptors (23). Ang-1-mediated signalling mechanisms involving Tie2 phosphorylation have the potential to maintain tight endothelial cell interactions (24). A reduction in vascular sprouting and the induction of tumour neovascularization can be achieved by blocking Tie/Ang-2 (11). In addition, the Ang-2/Tie signalling and VEGF signalling pathways synergistically promote tumour neo-angiogenesis, blockade of Ang-2 and VEGF delays tumour growth and enhances survival benefits through reprogramming of tumour-associated macrophages toward an antitumour phenotype as well as by pruning immature tumour vessels (25). Pericytes release Ang-1, which binds to endothelial cell TIE2 receptors, thereby efficiently inhibiting endothelial cell proliferation and promoting angiogenesis stability through downstream signalling pathways, such as Calpain, protein kinase B, and forkhead box O3A (26).

Anti-angiogenic targeted therapies inhibit the proliferation, invasion, and metastasis of malignant tumours by disrupting proangiogenic signalling between tumour cells and ECs ( Figure 3 ). Currently, the clinical application of AADs mainly includes antiangiogenic macromolecular monoclonal antibodies, small-molecule antiangiogenic tyrosine kinase inhibitors, and antiangiogenic-related molecule inhibitors. The most prevalent macromolecular monoclonal medicines are bevacizumab and ramucirumab, and the primary targets of these medications are VEGF-A and VEGFR. Bevacizumab as a first-line regimen for metastatic colorectal cancer (mCRC) has been shown in clinical trials to extend the median overall survival and progression-free survival of mCRC patients (27). However, due to its single targeting that is typically combined with chemotherapeutic drugs, bevacizumab, when combined with platinum-based therapies, has a considerable PFS advantage in metastatic non-small cell lung cancer (NSCLC) (28).Tyrosine kinase inhibitors, such as sorafenib, sunitinib, and pazopanib, control angiogenesis by preventing the function of receptor tyrosine kinases that accelerate angiogenesis. Since the approval of Sorafenib in 2007, significant progress has been made in the treatment of hepatocellular carcinoma (HCC) (29). Studies of combination therapy with molecule-targeted drugs and immune checkpoint inhibitors are underway, promising to benefit more patients (30). Currently, a Phase III trial has been conducted to effectively treat neovascular age-related macular degeneration (nAMD) by using faricimab, an antibody targeting Ang/Tie and VEGFR pathway (31).

The process by which tumour cells invade normal tissues and utilize the existing circulatory system to gather nutrition and oxygen is known as vascular co-option (32). Pezzella first discovered in 1997 that tumours could also grow in nonangiogenic ways in non-small cell lung cancer cells (33). Further studies have revealed that vascular co-option occurs in lung cancer, renal cell carcinoma (RCC), colon cancer, glioma and breast cancer (34–40). Under a light microscope, the characteristic pathological morphology of vascular co-option may be observed by staining tissue sections with the commonly used haematoxylin and eosin method (41). By quantitatively analysing CT and MRI images, angiogenic tumours can be distinguished from tumours that utilize vascular gain to accurately distinguish between these two phenotypes (41). Vascular co-option is characterized by intact vascular architecture of normal tissue within the tumour and tumour uptake of normal vessels surrounding the tissue rather than neovascularization, inflammation, or fibrosis (42). In contrast, tumour angiogenesis usually manifests as disorganization.

Angiogenesis is highly dynamic in time and space, with some tumour not only budding through vascular co-option but also utilizing sprouting angiogenesis to derive nutrients from them. There are two main ways for blood vessel absorption between tumour and adjacent tissues: (1) malignant cells replace normal epithelial cells (2) tumour cells invade the perivascular matrix (43). The molecules involved in the various stages of vascular co-option formation have been summarized by Ying Shao (44). The occurrence of vascular co-option seems to be related to the type of tumour and metastatic site. By analysing 164 cases of lung metastases (34), it was observed that 91.3% of breast cancer, 98.2% of colorectal cancer, and 62.3% of kidney cancer metastases exhibited some degree of vascular absorption growth. Additionally, 71.7% of breast cancer, 78.9% of colorectal cancer, and 37.7% of kidney cancer metastases had the vascular uptake growth pattern as their predominant pattern. Furthermore, evaluation of haematoxylin and eosin staining tissue sections from 45 cases of liver metastases found that a vascular resorption pattern was present in 96% of breast cancer liver metastases compared with only 32% of colon cancer metastases (45). Compared to renal and colorectal malignancies, vascular co-option growth patterns are more prevalent in breast cancer, and most colorectal cancer liver metastases induce a wound-healing response through angiogenesis. These data suggest that vascular co-option often occurs in breast cancer metastases to the lung and liver, which may help to explain why VEGFR antiangiogenic medication has had very little efficacy in treating metastatic breast cancer. The ability of tumour cells to resist such therapies through a mechanism known as vascular co-option is becoming increasingly obvious. Research by Victoria has demonstrated that although the drug sunitinib can inhibit the growth of subcutaneous xenografts, tumours that metastasize to the lung often utilize vascular co-option (34). Further evidence indicates that patients with HCC treated with sorafenib may experience a shift from vascular sprouting to vascular uptake driven by tumour cells (5). Additionally, intracranial glioblastoma tumours exhibit slower growth in response to anti-VEGF treatment but appear to adapt to angiogenesis inhibition by co-opting the host vasculature (6). The co-option of pre-existing brain vasculature contributes to malignant development without the production of sprouting angiogenesis, as further demonstrated by a model of malignant melanoma brain metastases (46).

Due to the lack of accurate marker analysis of vascular co-option, which can only be performed by analysing tissue slices of patients, studies on the mechanisms of vascular co-option are unclear. Current studies have shown that vascular co-option is related to the epithelial mesenchymal transformation (EMT) process and vascular adhesion of tumours. EMT refers to a phenotype in which ECs lose their polarity through a specific process to obtain a higher ability to invade and migrate (47). Tumour cells are more likely to break through the basement membrane of blood vessels and enter the blood vessels to absorb nutrients and oxygen. Therefore, the occurrence of vascular co-option in tumours is also correlated with the factors and signalling channels related to EMT process. As a known signalling pathway in EMT process, the Wnt signalling pathway is involved in tumour EMT process and plays a crucial role in the occurrence and development of tumours (48). In glioma, treating cells with LGK974, an inhibitor of the Wnt signalling pathway, significantly reduced Wnt7a expression in vivo, preventing tumour cell migration along vascular endothelial cells and contact with blood vessels (36). Serine proteinase 2 (35), runt-related transcription factor-1/actin-related protein 2/3 complex (49), fibroblast activation protein α (50) are signalling molecules involved in EMT. The deletion of these genes will lead to changes in cell motility and promote tumour selection of vascular co-option mode mediating the malignant development of cancer.

Another widely studied change in vascular co-option is vascular adhesion. L1CAM, as a cell adhesion factor, is involved in the development of the nervous system and the progression of malignant tumours (51). The dynamic nature of L1CAM adhesion interactions may be particularly beneficial for cancer cells seeking cooperative vasculature when invading tissues, mediating and participating in vascular co-option of BMS cancer cells (35). L1CAM depletion significantly reduced the ability of H2030-BrM3 and MDA231-BrM2 cells to spread on the surface of the capillary cavity and decreased the activity of brain metastasis of cancer cells, thereby mediating co-option and metastatic growth of brain capillaries. Vascular co-option is prevented by reducing tumour cell EMT progression or targeting tumour cells’ propensity to adhere to blood vessels ( Figure 4A ).

In 1990, Burri discovered that capillary beds expanded by generating elongated columns of intravascular tissue in the lungs of rats and named this development pattern intussusceptive microvascular growth (IMG) (52). Since then, IMGs have been found in both embryonic and adult tissues and organs (53, 54). As research has progressed, IMGs have also been found to be involved in the abnormal proliferation of tumour vessels. Patan demonstrated that intussusceptive obstruction is an important mechanism of tumour angiogenesis by examining the growth of human colon adenocarcinoma (LS174T) (55). This is a mechanism that also occurs in several other tumours, such as colon (55, 56) and mammary carcinomas (57), melanoma (58), B-cell non-Hodgkin’s lymphoma (59), and glioma (60).

According to Ribatti, the steps in the development of intussusception are summarized as involving contact between neighbouring endothelial cell walls, reorganization of connections between ECs, central perforation between interconnected bilayers of ECs to form a core of mesenchymal pillars, and extension and invasion of the pillars by myofibroblasts and pericytes (8). Finally, the pillar increases in diameter, forming collagen fibres and, eventually, a capillary network ( Figure 4B ). Thus, in contrast to tumour sprouting angiogenesis, the most distinctive feature of intussusception is the absence of endothelial cell proliferation and the formation of ‘transvascular pillars’.

The mechanism underlying the onset and development of intussusceptive angiogenesis (IAs) is complex and dynamic, and no definitive studies exist that prove its development mechanism. Nevertheless, researchers have identified several factors that play key roles in this intricate process. First, shear stress generated by blood flow in the vasculature is an important aspect of IA development. Djonov demonstrated changes in branch morphology through blood pressure and blood clamping of a dichotomous branch in the developing chorioallantoic membrane (CAM) of the chick embryo (61). Second, VEGF-A is involved in SA and vascular co-option of tumour angiogenesis, and a considerable body of research indicates that the development of IA is also linked to VEGF (10). It is thought that shear stress triggers VEGF expression in endothelial cells adjacent to myofibers through the diffusion of nitric oxide in the vascular microenvironment, promoting tumour angiogenesis (62, 63). There is a correlation between the progression of intestinal obstruction and the production of VEGF in CAM micro-vessels. Inhibition of VEGF signalling pathway inhibits the development of capillaries that are dependent on intestinal obstruction (64). Furthermore, throughout the IA process, researchers have discovered that FGF2 increases the development of intraluminal pillars in ECs and pericytes (65). Observations have also shown that vascular morphological alterations through intussusceptive angiogenesis occur in hepatocellular carcinoma models during treatment with sirolimus, an inhibitor of the mammalian target of rapamycin (mTOR) pathway (7). The shift in angiogenesis from sprouting to intussusceptive might be an adaptive reaction to therapy using various antitumour and antiangiogenic drugs (8).

In vitro, researchers demonstrated the reproduction of the structured vascular channels present in human tumour tissue was shown in 1999, even in the absence of Ecs, in highly invasive melanoma cells (66). They named this mechanism vasculogenic mimicry. VM has been observed in numerous malignant tumours in recent years, including melanoma (67, 68), glioblastoma (69), HCC (70, 71), breast cancer (72, 73), and lung cancer (74). Unlike conventional angiogenesis, VM does not involve ECs and is characterized by tumour cells constituting vascular channels and the dense deposition of extracellular matrix (75). The presence of CD31/CD34-negative and PAS-positive cells and erythrocytes within the vasculature is considered a hallmark of VM (76). Saber provided microscopic and immunohistological data to facilitate the identification and understanding of in vitro and in vivo VM processes (77).

As previously mentioned in this review, hypoxia is a significant driver that not only affects tumour angiogenesis but also plays an essential role in antiangiogenic therapy (78, 79). HIF-1 binds to the HRE on the promoter sequences of Twist, Snail, and ZEB2 in the nucleus to regulate their expression. HIF-1 enhances LOXL2 expression in hepatocellular carcinoma, which promotes extracellular matrix remodelling and VM via the Snail/FBP1/VEGF pathway (80). A hypoxic microenvironment was observed to enhance VM in a melanoma mouse model compared to controls and was positively correlated with HIF-1α and HIF-2α (81). HIF-1 also controls the expression of other VM-related molecules, including VEGF, Twist, LOX, and matrix metalloproteinase (MMP). The expression of VEGF in melanoma has been reported to promote VM development by activating the PI3K/AKT pathway (82).

Knockdown of VEGF also reduces the expression of MMP and vascular endothelial-cadherin(VE-cadherin) (83). VE-cadherin erythropoietin-producing hepatocellular receptor A2 (Eph A2) plays an important role in VM. VE-cadherin regulates the phosphorylation level and localization of Eph A2, thereby activating focal adhesion kinase (FAK) and extracellular signal-regulated kinase (ERK) 1/2, and subsequently, activated Eph A2 can also activate the PI3K signalling pathway directly without relying on FAK and ERK1/2 (84). Furthermore, PI3K can upregulate MMP expression, which in turn activates MMP2, ultimately leading to the cleavage of laminin 5γ2 into the γ2’ and γ2x fragments, promoting extracellular matrix remodelling and VM formation (85, 86).

Considerable progress has been made in inhibiting tumour angiogenesis by targeting the signalling pathways involved in VM ( Figure 5 ). Ginsenoside Rg3 decreased the expression of VE-cadherin, EphA2, MMP9, and MMP2 to successfully block pancreatic cancer mimic angiogenesis (87). As a novel candidate for antitumour VM and anti-covariance treatment, miR-27b might bind to the 3’-untranslated region (3’UTR) of VE-cadherin mRNA to reduce the production of VE-cadherin in ovarian cancer cells (88). By suppressing STAT3 and PI3K/AKT, curcumin decreases the ability of hepatocellular carcinoma cells to simulate angiogenesis (89). By disrupting the PI3K/MMPs/Ln-52 signalling pathway, NCTD prevents tumour development and VM in human GBCs both in vitro and in vivo (90).

BMDCs play a crucial role in the pathogenesis and dissemination of malignancies due to their ability to suppress the immune system and promote the migration and vascularization of tumour cells. Comprising various cell types such as lymphocytes, stromal cells, bone marrow cells, and endothelial progenitor cells, BMDCs exhibit heterogeneity. They are recruited around blood vessels to actively participate in, either directly or indirectly through the secretion of signalling molecules ( Figure 4C ). Several haematopoietic populations from the bone marrow have been reported to contribute to tumour angiogenesis, such as monocytes (91), macrophages (92), granulocytes, and neutrophils (93, 94).

Daniel and his colleagues (95) used genetically labelled bone marrow progenitor cells and demonstrated, using high-resolution microscopy and flow cytometry that early-stage tumours recruit bone marrow-derived endothelial progenitor cells These endothelial progenitor cells differentiate into mature ECs, integrating into the developing tumour vasculature, thereby promoting tumour growth and progression. However, the mechanism by which tumours recruit BMDCs to the site of angiogenesis remains unknown. Tumours enhance the recruitment of BMDCs to the site of angiogenesis by secreting integrin avb3, a receptor that mediates cell−cell and cell–extracellular matrix adhesion (96). Platelets further mobilize and recruit CXCR4+ bone marrow-derived cells to promote vascularization (97). Tumours also induce myeloid cells to secrete factors associated with vascularization to promote tumour angiogenesis and progression. MMP-9, which controls the switch for angiogenesis by releasing VEGF from its membrane-bound form, has been shown to be expressed in myeloid cells (98). Rose demonstrated that HIF-1α partially induced the recruitment of bone marrow-derived CD45+ myeloid cells with Tie2+, VEGFR1+, CD11b+, and F4/80+ subpopulations, along with endothelial and pericyte progenitor cells, by increasing SDF1α and promoting the neovascularization of glioblastoma (99). Additionally, HIF-mediated negative regulation of PHD2 mobilizes BMDCs to regulate angiogenesis (100).

The recruitment of myeloid cells from bone marrow is directly associated with the emergence of tumour resistance or recurrence after antiangiogenic treatment. In response to antiangiogenic treatment, it has been demonstrated that macrophages increase the production of a number of angiogenic molecules, including FGF-1, FGF-2, MMP9, and Ang2 (101–104). Partially overcoming tumour resistance may be achieved by combining anti-VEGF with monotherapy targeting CD11b+Gr1+ myeloid cells (105). Thus, effective ablation of myeloid cells’ proangiogenic capacity may be achieved by extensively targeting them in cancer therapy (3). Inhibiting the gamma isoform of PI3K (PI3Kγ), which is largely expressed in myeloid cells and maintains their immunosuppressive and proangiogenic functions, maybe a promising targeting approach (106).

Cancer stem cells (CSCs) represent a small subset of cells within solid tumours capable of self-renewal and differentiation into various cell types present in the tumour. They play a crucial role in tumour metastasis, recurrence, and resistance to chemo and radiotherapy across a spectrum of cancers (107). Recent research has highlighted the role of CSCs in promoting angiogenesis through the release of angiogenic factors and exosomes (108–110). For instance, For example, CSCs in breast cancer have been found to secrete stromal cell-derived factor-1 and VEGF to facilitate angiogenesis (111). Further research discovered that enhanced IL-8 secretion by CD133+ CSCs in HCC not only induced tumour angiogenesis but also boosted tumour self-renewal and initiation (107). Additionally, emerging evidence suggests that tumour cells possess the ability to differentiate into phenotypes of ECs that are engaged in tumour angiogenesis. According to research on glioblastoma, ECs in varying quantities (range 20–90%, mean 60.7%) contain the same genetic abnormalities as tumour cells. This finding raises the possibility that GBM-derived stem cells can differentiate into ECs in vitro (112). This result suggests that tumour cells may have CSC origins, providing insight into the formation of VM.Studies by Sun et al. have shown that ALDH1+ and CD133+ breast cancer cells produced more VE-cadherin and generated VM channels (113). To support the formation of CSCs and preserve their stem properties, vascular niches in the tumour microenvironment also release growth factors through autocrine and paracrine pathways (107). Tumour growth is further encouraged by the positive feedback loop between tumour stem cells and angiogenesis.

From a therapeutic perspective, targeting CSCs is essential to impede tumour progression. The conventional approach involves blocking aberrant signalling pathways in CSCs, mainly including the Wnt, Nuclear factor-κB (NF-κB), Notch, Hedgehog, JAK/STAT, PI3K/AKT/mTOR, TGF/SMAD, and PPAR pathways. Napabucasin, a small-molecule inhibitor of STAT3, has shown efficacy in preventing the spread of biliary tract cancer, glioblastoma, and small-cell lung cancer (114–116). Clinical trials combining napabucasin with paclitaxel for gastric cancer have demonstrated favourable pharmacokinetic profiles, safety, and patient tolerability (117). Additionally, LGR5, a seven-transmembrane receptor from the G protein-coupled receptor family connected to the Wnt pathway that contains leucine-rich repeats, was first discovered as a marker of intestinal stem cells (118). LGR5 is a unique functional marker and a prognostic indicator that, by stimulating the Wnt pathway, can drive EMT and become a potential therapeutic target (119). In colon cancer, targeting LGR5 using antibody-drug conjugates greatly reduced tumour size and proliferation (120). Since normal human stem cells and tumour stem cells usually express the same surface and protein markers, therapeutic medication development faces significant challenges. Table 1 summarizes some of the signature markers of tumour stem cells for reference. To boost antitumour efficacy, there is rising interest in combining immunotherapy with conventional chemotherapy and targeted medicines. The most promising immunotherapy is chimeric antigen receptor (CAR) T-cell treatment, which has been licensed for B-cell acute lymphoblastic leukaemia and B-cell lymphoma (144). CAR T cells targeting cd133 in glioma not only have good efficacy in xenograft models but also have no negative effect on normal human CD133+ haematopoietic stem cells (137). As a result, CART133 cells may be a useful treatment option for CSCs in different solid cancers.