Circulating tumor cells are very heterogeneous. CTC have different levels of malignancy and capacity to found metastases. On the clinical side, this means that if we would like to use tumor cells in blood to assess more precisely the patient’s risk of developing metastases, we would need to quantify, in addition to the number of tumor cells collected from blood very sensitively without bias and counted diagnostically, the number of those cancer cells in transition from epithelial to mesenchymal phenotype (EMT) [
16], those with full mesenchymal phenotype, those with “stem” characteristics, and the number of circulating tumor microemboli [
8,
16]. Epithelial-to-mesenchymal transition (EMT) of carcinoma cells results in reduced expression of epithelial markers and increasing expression of mesenchymal markers, such as vimentin, also expressed in leukocytes, and N-cadherin. EMT and CTC heterogeneity represent a strong limitation for methods aiming at isolating CTC using antibodies since specific markers of tumor cells, i.e. markers expressed by all the tumor cells and not expressed at all by other blood cells or circulating non-tumor cells, are unknown at present [
16].
Besides N-cadherin and vimentin, markers of EMT include nuclear localization of β-catenin, and increased expression of transcription factors such as SNAIL, SLUG, TWIST, ZEB1, ZEB2 and/or TCF3 inhibiting the production of E-cadherin. EMT is associated with an increased cell capacity for migration and invasion, as well as resistance to anoikis and apoptosis [
18,
19]. Different subsets of circulating tumor cells may have a range of intermediate phenotypes, between epithelial and mesenchymal, which could be a sign of high plasticity and “stem” character [
20,
21].
. This plasticity is thought to be linked to genetic and epigenetic changes of cancer cells [
22].
Cancer cells with a stem phenotype (CD44
+CD24
−/low, ALDH1
+) may also circulate in blood [
21,
23]. Some researchers have recently been able to obtain ex vivo culture of CTC from a minority of breast cancer patients [
24]. The possibility to consistently culture CTC from cancer patients would provide an extraordinary tool to test in vitro their drug sensitivity as currently performed with leukemic cells. However, the majority of CTC are expected to be out of cycle [
13], not susceptible to be cultured or very resistant to re-enter the cycle. Therefore, the clinical utility of this approach is linked to the development of an assay which would 1) consistently stimulate CTC proliferation in vitro; 2) assess if the proliferating cells are really tumor cells (as circulating non-tumor cells precursors are highly susceptible to proliferate); 3) assess whether the method which stimulates CTC proliferation artificially affects, or not, CTC responsiveness to drugs.
Some authors have demonstrated the existence of “tissue-competent” CTC for brain metastases (EPCAM
− expressing HER2, EGFR, NOTCH1 and HPSE), i.e. « soil-specific » seeds, opening the way to the exciting perspective of future tests predicting the type of upcoming metastasis [
25]. Results obtained in animals [
26] have shown that “stem tumor cells” are highly concentrated in the blood compartment as compared to the primary tumor site. If studies carried out in patients could confirm that blood contains the most malignant tumor cells genotypes, the clinical interest of using the blood tumor cells compartment for theranostic analyses would be huge. In fact, the study of different samples from primary tumor tissues has shown extensive intra-tumor genetic heterogeneity [
27,
28], which could lead to biased theranostic tests performed on localized tissue sampling missing the most malignant genotypes. Cancer stem cells have also been described to undergo bidirectional transformation to a differentiated phenotype and vice versa [
29], a character that could be shared by circulating cancer stem cells and make even more difficult assessing their metastatic potential.
Another expression of CTC heterogeneity and malignancy is the possible presence in blood of circulating tumor microemboli (CTM). Originally described in mice studies [
30], then in humans [
8,
31‐
33], CTM are known to have an increased metastatic capacity. They may include stromal cells and bring “their own soil” [
34]. Recently, it has been shown in animal studies that they arise from oligoclonal tumor cells groupings and not from intravascular aggregation events [
35]. Finally, recent studies using a mouse model of pancreatic cancer and single-cell RNA sequencing have identified the expression of extracellular matrix genes, including SPARC, in CTC, an interesting finding that the Authors confirmed in CTC from patients with pancreatic cancer [
36]. Such data demonstrate that CTC may be characterized by a very high level of plasticity, which is presumably related to their capacity of creating distant metastases. These and further studies focused on heterogeneity of tumor cells in blood should be able to shed more light on the markers associated with their increased metastatic potential.
From a technical point of view, CTC heterogeneity represents a challenge for CTC tests developers. In fact, the optimal CTC test would require the isolation of all types of CTC without any loss allowing to further perform their immune-molecular characterization. CTC heterogeneity also highlights the difficulty of using antibodies or cocktails of antibodies to isolate/identify all types of CTC as we lack specific markers expressed in all types of CTC and not expressed in non-tumor cells [
16]. There is a need for a broad-spectrum specific cocktail of cell surface epithelial and mesenchymal markers [
19] covering all potential CTC phenotypes. This cocktail, however, could increase the chance that at least some of these markers cross react with, or are expressed by, blood cells and/or other circulating non-tumor cells, which would lead to false-positive results [
16]. Concerning tumor-specific markers, melanoma-associated antigens (MAGE) are specifically expressed in melanoma cells. Still, they are generally used as transcripts in RT-PCR methods to detect CTC, which do not allow counting of the CTC number nor CTC immunomolecular characterization [
37,
38]. Tissue-specific markers, such as prostate-specific antigen (PSA) for prostate cancer and mammaglobin for breast cancer, are not tumor-specific markers as they can be expressed in non-tumor circulating epithelial cells [
39] and be down regulated during dedifferentiation of tumor cells [
16]. Some markers, like HER2 and EGFR are expressed at higher levels in cancer cells as compared with normal cells in certain tumor types [
9]. Since they are not expressed in all the tumor cells, these markers can be used to characterize CTC and to guide the use of targeted therapies, but are not useful for the systematic diagnosis of CTC. In fact, in metastatic breast cancers, HER2 positivity rates of CTC vary widely, between 27 and 63 %, depending on CTC isolation and characterization methods. Furthermore, up to 49 % of patients with HER2 negative primary tumors have HER2 positive CTCs and conversely, up to 77 % of patients with HER2 positive primary tumors have HER2 negative CTCs [
9].
To summarize, CTC heterogeneity is a biological characteristic of CTC and a challenge for CTC tests developers, who have difficulties finding CTC-specific markers. For CTC, their heterogeneity could limit their metastatic potential in the blood microenvironment, for instance by “diluting” the effect of platelets [
40,
41] which are know to act as protection against immune-mediated lysis [
42] and as EMT inducers [
43].