CTCs arise from the tumor cells of primary tumors, metastases, lymph nodes or disseminated tumor cell pools (5,8). Before tumor cells enter the circulating system, a series of phenotypical and genotypical changes occur, including cytoskeletal reorganization, protease secretion, changes to the expression profile of adhesive proteins and receptors, and these changes allow CTCs to acquire an invasive phenotype and detach from the primary tumor (Fig. 1A) (9,10). Epithelial-mesenchymal transition (EMT) serves a key role in the loss of cell polarity and decrease of cell-cell adhesion resulting in an increase in the invasiveness of tumor cells (11,12). Predisposing factors of EMT include transforming growth factor-β (TGF-β), interleukin-35 and interleukin-6 which are secreted by macrophages, activation of the WNT signaling pathway and platelet-derived growth factor, and exposure to nicotine, alcohol and ultraviolet light also exacerbate this process (12–17). Epidermal growth factor (EGF) and hepatocyte growth factor, which are secreted by monocytes and neutrophils, and the presence of inflammation are also strong predisposing factors for EMT (18). Additionally, Twist, SNAIL, Zeb and other genes have also been shown to serve a crucial regulatory role in EMT (19–21). As tumor cells undergo EMT, the number of intercellular junctions reduce notably, as well as the expression of epithelial markers, such as keratin, E-cadherin and epithelial cell adhesion molecule (Ep-CAM) (22). Conversely, expression of mesenchymal markers such as vimentin are increased, and subsequently the morphology of the cells are altered allowing for detachment from the primary tumor, and ultimately becoming invasive (14,23). Centrosome amplification and CTC clusters are also hypothesized to confer invasive and metastatic properties to tumor cells (24,25).

Once cells detach from the primary tumor, the tumor cells near the neovascular site enter blood vessels through the sparse vascular endothelium under increased pressure of tumor tissue growth which facilitates extrusion (26). Additionally, changes in the extracellular matrix increase the synthesis of key components such as collagen and fibronectin, and simultaneously promote the development of abnormal structures in the extracellular matrix, through increased activity of members of the cathepsin and matrix metalloproteinases families of proteins, which are secreted by neutrophils and mast cells (27,28). Macrophages and inflammatory monocytes enhance the migration of tumor cells by increasing the synthesis and cross-linking of collagen and fibronectin (29,30). Natural killer cells reduce the expression of fibronectin by secreting interferon-γ, thereby limiting the migration of tumor cells (31). These changes all contribute to the breakdown of the extracellular matrix and intravasation of tumor cells into blood vessels. Subsequently, tumor cells use chemotaxis to migrate to sites of high vascular permeability to intravasate (Fig. 1A), and macrophages serve an essential role in this process. Macrophages alter vascular structures and increase permeability by secreting a series of growth factors such as the vascular endothelial growth factor (VEGF), fibroblast growth factor, and placenta-derived growth factor, which facilitate migration of tumor cells through the vascular endothelium (32,33). Additionally, macrophages recruit tumor cells via the epidermal growth factor (EGF) paracrine loop to sites where vascular permeability is higher and thus more readily penetrable (34). This paracrine loop is regulated by the secretion of colony-stimulating factor (CSF-1) from tumor cells. CSF-1 regulates macrophages, increasing the production of EGF, and these macrophages are recruited to blood vessels, forming a chemical concentration gradient of EGF to guide the tumor cells towards the blood vessels (35,36). This macrophage-mediated chemotaxis can be antagonized by T cells (37).

CTCs in the blood face three challenges to survive, blood flow shear, anoikis, and immune cell identification and killing (38). Blood flow shear force is a double-edged sword. Under certain conditions, blood flow shear can increase the invasive and metastatic capacity of CTCs by activating yes-associated protein 1, as well as the ability to adhere to and penetrate blood vessels (39). Conversely, CTCs can be physically destroyed by blood flow shear (40), and the longer CTCs are present in the blood circulation, the lower their proliferative capacity becomes, and eventually the cells enter dormancy (41). Tumor cells undergoing EMT exhibit an improved ability to resist the blood flow shear force (42). When the tumor cells detach from their original microenvironment, the original extracellular matrix is unable to provide the necessary cytokines and signals for growth and survival resulting in death. This programmed death of the tumor cells is called anoikis. In CTCs, activation of the Akt pathway triggers tropomyosin-related kinase B and inhibits caspase-related apoptosis to counter anoikis (43). Furthermore, the Akt signaling pathway is activated in tumor cells and resists apoptosis induced by tumor necrosis factor-related apoptosis-inducing ligand, and this is mediated by vascular cell adhesion factor-1 (VCAM-1) (44). Platelets serve a crucial role in the resistance of CTCs to immune cell-mediated killing (Fig. 1B). Platelets can adhere to CTCs, forming an armor to resist blood flow shear (45), and natural killer (NK) cells recognize and kill the CTCs (46,47). However, platelets induce EMT in CTCs via the NF-κB and TGF-β signaling pathways (48), and as mentioned earlier, CTCs which have undergone EMT are more resistant to blood flow shear (42). Platelets may also prevent CTCs from NK cell recognition by transferring major histocompatibility complex 1 molecules to CTCs and secreting TGF-β to decrease the expression of NK cell NK Group 2D (49,50). Furthermore, platelets act as a link between CTCs and CD11⁺ macrophages to form cell clusters which help CTCs survive in the bloodstream (51). Additionally, CTCs selectively express Programmed death-ligand 1 (PD-L1), CD47 and Fas/Fasl which reduce the elimination of CTCs by immune cells (52–55).

Extravasation of CTCs from blood vessels is associated with several factors, such as the vascular endothelium linkage state, blood circulation state, capillary structure and adhesion of CTCs to the vessel wall (56–59). When CTCs pass through the capillaries, the flow rate decreases and CTCs adhere to the vascular endothelium with the assistance of platelets and neutrophils (60,61). Mononuclear macrophages are recruited to the vascular endothelium, and CTCs adhere through C-C Motif Chemokine Ligand 2, and subsequently secrete VEGF, increasing the permeability of vascular endothelium and thus assisting extravasation of CTCs (62,63).

CTCs do not immediately contribute to metastasis, but remain dormant possibly for several years following extravasation and colonization of a distant site (Fig. 1B) (64). However, the processes underlying the ability of tumor cells to survive and enter a dormant state are complex. Myeloid cells, monocytes and macrophages, including neutrophils and the factors secreted by these cells accumulate at the metastatic site, producing an inhibitory environment for immune responses, thus assisting the tumor cells entry into dormancy prior to the arrival of tumor cells at the dissemination site (34,65,66). Additionally, tumor cells resist apoptosis through activation of the Akt signaling pathway and thus improving tumor viability (4,44,67). Tumor cells enter dormancy and are divided into three stages, angiogenesis dormancy, immune dormancy and tumor cell dormancy (64). Tumor growth requires a large number of new blood vessels to ensure a sufficient supply of nutrients (68,69). During the dormant phase, angiogenesis is not activated (64). Proliferating tumor cells are in equilibrium with tumor cells that die due to a lack of oxygen and other nutrients due to the angiogenesis dormancy (68,70). In terms of immune dormancy, it has been confirmed in animal models that the presence of lymphocytes, particularly CD8⁺ T cells, can maintain tumor dormancy and reduce metastasis (71). The deletion of CD8⁺ T cells results in a 100% tumor metastasis rate (72). In addition, EMT in tumor cells (26), and disruption of the balance between extracellular signal-regulated kinases and p38 mitogen-activated protein kinases (73), alters the expression of urokinase plasminogen activator receptor and integrin-β1 (74), activation of TGF-β2 signaling pathway (75) and expression changes in Myc gene (64), which all regulate the entry of tumor cells into a dormant state. When tumor cells enter dormancy, immune cells and conventional cytotoxic chemotherapy does not kill these cells; in conditions of post-operative stress, such as ischemia-reperfusion, sympathetic excitation, inflammation and hypercoagulability, these dormant cells may be activated, resulting in recurrence and metastasis in patients who underwent primary tumor resection (34,76). Exit from dormancy is also a complicated process. Tumor cells exit dormancy and gain proliferative capacity, when they undergo mesenchymal-epithelial transition (26), and this process involves tumor necrosis factor-α, interleukin-6 (77), integrin, activation of the Fak signaling pathway (78) and prostaglandin E2 (79). TGF-β receptor antagonists awaken tumor cells from dormancy by blocking the bone morphogenetic protein signaling pathway and thus improving tumor stem cell characteristics (80).

After CTCs have been isolated using one of the above methods, confirmation of their identity needs to be verified, and a series of tests are performed to assess the cells and obtain information regarding the original tumor information from which the CTCs were derived (Fig. 2B). Immunocytochemistry is used to determine the expression of specific markers, thereby defining the identity of the CTCs. For example, in CellSearch, cells ≥4 µm, 4,6-diamino-2-phenylindole (DAPI⁺), keratin⁺ and CD45⁻ are used as criteria for determining CTCs (115), and the expression of molecules on the surface of CTCs may also be analyzed by this technique, such as by staining for EMT-associated molecules (116). The classification of CTCs may also be confirmed using a 4-color fluorescence detection system (112). If cells are isolated and cultured, the proteins secreted by the cells can be detected using EPISPOT, to determine the identity of the CTCs (117). PCR can also be used for the identification of CTCs and for profiling the transcript of tumors, EMT, stem cell characteristics such as Ep-CAM, aldehyde dehydrogenase 1 and Twist1 (118). With the emergence of whole-genome amplification (119), whole-genome bisulfite sequencing (120), whole-genome and exome single-cell sequencing (121), next-generation sequencing (117), third-generation sequencing (120) and low-pass whole genome and targeted next-generation sequencing (121), our understanding of CTC gene mutations has increased notably. In addition to the single-cell level, the association between the CTC phenotype and gene mutations can be visualized using fluorescence in situ hybridization (FISH) (122), without the need to separate individual cells by micromanipulation following CTC enrichment. For RNA, the emergence of technologies such as single-cell RNA sequencing (8) and digital droplet PCR (123), have allowed for a deeper understanding of the biological changes of CTCs at the transcriptional level. Furthermore, other novel technologies allow for assessment of the changes in protein expression levels from an individual CTC. For example, single CTC resolution western blot analysis can quantify 12 proteins simultaneously, this technology, when combined with flow cytometry, can quantify >40 proteins, significantly improving screening efficiency (124,125). Fluctuations in pH or detection of lactate concentrations have been employed in a microfluidic device to identify the metabolic status of CTCs (126).

Due to the rarity of CTCs, and the difficulty in isolating CTCs whilst simultaneously maintaining their survival outside of the body, certain studies are performed at the cellular or animal level compared with clinical studies of CTCs. In the case of patients with relatively high concentrations of CTCs, cell and animal research of their CTCs is possible (Fig. 3A). In a patient with colorectal cancer, the CTC concentration was >300/7.5 ml, thus blood was enriched and cultured, and a cell line based on their CTCs was successfully obtained (127). Studies on gene transcription, protein expression and secretion levels demonstrated that CTCs have primitive tumor cell and stem cell characteristics, mixed epithelial and mesenchymal phenotypes, and strong tumorigenic ability in animal experiments. In breast cancer, CTCs obtained from patients have mutations in the gene of PIK3CA, estrogen receptor 1, fibroblast growth factor receptor 2, and drug sensitivity to the mutations (128). Another study showed that CTCs with human epidermal growth factor receptor 2/epidermal growth factor receptor/Heparanase/Notch1⁺ have a tendency to metastasize to the brain in animal experiments (129). Similar studies have also been performed in prostate cancer (135), non-small cell lung cancer (136), and small cell lung cancer (137,138), and the tumorigenicity, drug resistance, and mutation sites of CTCs having been confirmed. Through these studies, a more comprehensive understanding of the relationship between CTCs and the metastatic tendency of different organs can be determined (139), and this individualized treatment (128) and the mechanisms underlying increased invasion and metastasis may be elucidated (140).

In clinical studies, research can be broadly divided into two categories, bivariate studies, and multivariate studies. In bivariate studies, the two variables are CTC-related data and patient clinical follow-up data (Fig. 3B). CTC-related data includes the number of CTCs, and protein, DNA and RNA expression profiles. Patient follow-up data primarily include disease-free survival, overall survival, chemotherapy medication and methods, and recurrence. CTCs are used as a predictor of clinical prognosis, chemoresistance, and for detection of tumor metastasis and recurrence. For example, in advanced pancreatic ductal adenocarcinoma, the number of CTCs is an important predictor of disease-free survival and overall survival prior to, and following first-line chemotherapy; the higher the number of CTCs, the shorter the disease-free survival and overall survival of patients (123). However, the mRNA expression levels of activated leukocyte cell adhesion molecule, POU Class 5 Homeobox 1B and smoothened are increased in CTCs, and this is associated with less favorable overall and disease-free survival (123). In metastatic castration-resistant prostate cancer patients, patients with CTCs expressing nuclear localization androgen receptor splice variant 7 (ARV7) exhibit improved responses to taxanes and androgen receptor inhibitors; conversely androgen receptor inhibitors result in improved outcomes compared with taxanes in patients whose CTCs do not express nuclear localization of ARV7 (133). In breast cancer, CTCs expressing Notch signaling pathway markers are closely associated with brain metastasis (132). Related bivariate studies have also been performed on non-small cell lung cancer (131), colorectal cancer (141), liver cancer (130) and kidney cancer (116). However, studies on hepatic carcinoma have rarely yielded a negative result, counts of CTCs and types of EMT are not associated with clinical stage and predictive recurrence of hepatic carcinoma (130). In addition, CTCs and solid tumor DNA (142) or protein analysis (143) were used as two elements for correlation analysis to compare the variation of primary tumors and CTCs. However, in patients with lung cancer, there was no correlation between PD-L1 on CTCs and in tumor tissues (106).

In multivariate studies, in addition to the aforementioned CTC data and patient follow-up data, TNM staging (144) or other liquid biopsy markers, such as circulating tumor DNA (145) and extracellular vesicles (146), can be used together as predictors of prognosis to improve prediction accuracy. For example, in colorectal cancer, postoperative CTC counts are more valuable than preoperative counts, and TNM staging, and postoperative CA724 and CTCs counts are more accurate for predicting disease-free survival of patients, and there is a model to predict early recurrence and postoperative survival rate using postoperative CA724, CTCs counts, which can include or exclude TNM staging (141). As mentioned above, CTCs are closely associated with platelets and immune cells in the circulatory system; therefore, by combining CTC data, immune-inflammatory cell counts in circulating blood, coagulation status and other factors with patient follow-up data, the accuracy of predicting prognosis and assessing risk can be improved. In metastatic breast cancer, the ratio of CTCs to blood inflammatory cells, such as the ratio of CTCs to monocytes and lymphocytes, can be used as a predictor of prognosis (147). Also in metastatic breast cancer, by combining CTC data with thrombin-anti-thrombin III, fibrinogen, D-dimer and patient follow-up data, it has been shown that the hypercoagulable state contributes to tumor cell metastasis (148).