Heterogeneity in Circulating Tumor Cells: The Relevance of the Stem-Cell Subset

1Despite recent medical advances, metastasis, tumor relapse, a lack of effective therapies, and drug resistance remain the major causes of death for tumor patients. Research on circulating tumor cells (CTCs), which have been detected in the majority of epithelial cancers, is a dynamic field of translational and basic research, with more than 270 clinical trials having evaluated CTC enumeration as a biomarker. Further, several studies on CTCs aimed to understand critical pathways that mediate cancer dissemination, and may not be apparent in primary or metastatic tumors.

2Cancer metastasis occurs when tumor cells dissociate from a primary tumor and migrate to distant organs through the peripheral vasculature. Cancer cells within the primary tumor lose cell-cell adhesion and acquire features of invasiveness and motility due to epithelial-to-mesenchymal transition (EMT), leading to their intravasation and survival in blood or lymph vessels, and finally their extravasation at distant sites; here they are subjected to mesenchymal-to-epithelial transition (MET), and proliferate to establish metastatic lesions. Whether the release of CTCs into the circulation is a predetermined biological process still remains a debatable matter. Cancer cells can enter circulation long before a tumor is diagnosed; the majority of cells die and only a minor fraction contains viable metastatic precursors that infiltrate organs and survive as disseminated seeds for eventual relapse. Therefore, metastatic colonization might be a highly inefficient process, but once it is established, current treatments generally fail to provide persistent responses.

3Tumors are thought to arise from mutant normal stem cells in their native niches, or from the progeny of cells that retain their tumor-initiating capacity. Once at distant sites, these cells survive and proliferate through interactions with specialized niches. Data from a number of studies demonstrate that CTCs with stemness properties exist and represent the most aggressive tumor cells in circulation.

4In this review, we will discuss current concepts and will address questions relative to the heterogeneity of CTCs and the presence of cancer stem cells circulating in the bloodstream (See Table 1).

5CTCs are very rare cancer cells released from a primary tumor. A lot of evidence documents that CTCs represent a heterogeneous pool of tumor cells, with highly variable, but generally short, survival times; nevertheless, the pool of these cells in the bloodstream can be continuously replenished by the release of replicating tumor cells in metastatic foci, and thus relates to cancer dormancy. To date, relatively little is known regarding the number of CTCs in cancers, their molecular and biological heterogeneity, and their functionalities. Although heterogeneity in CTCs has not yet been fully defined, a fraction of these cells are thought to be viable metastatic precursors capable of initiating a clonal metastatic lesion. A subset of CTCs is represented by inactive non-cycling cells, termed dormant cancer cells, which act as latent tumor-initiating seeds and eventually reawaken, and might not respond to chemotherapeutics used in clinics. By uncovering phenotypes of CTCs, CTC heterogeneity would be dissected in relation to metastatic competence.

6Recently, clusters of CTCs, consisting of 2–50 cells, have been detected in the vasculature of patients with cancers of different origins, which arise from oligoclonal tumor cell groupings, including breast, prostate, pancreatic, glioblastoma, head and neck cancers. Cells within micro-clusters might be protected either from anoikis, associated with the loss of cell-cell adhesion and attachment to the basement membrane, or shear stresses in the circulation. Both the structural deformability of aggregated cells and the presence of vascular shunts may allow CTC-clusters to circulate, while inducing a rapid clearance within distal tissues. Although rare, CTC-clusters present a dramatically enhanced metastatic potential, as confirmed in mouse models, and by the adverse prognosis in patients with high numbers of CTC-clusters. Additionally, neutrophils directly interact with CTCs, defining CTC–neutrophil clusters, and support cell cycle progression in circulation. This interaction in turn leads to more efficient metastasis generation and represents key vulnerabilities of the metastatic process for drug targeting in breast cancer. Mesenchymal transcripts were found to be expressed in tumor cells from CTC-clusters in a human breast tumor, conferring both migratory and stem-like properties. A recent study has suggested that the coexpression of EpCAM, CD44, CD47, and MET in CTCs identifies a subset of cells with increased metastatic capacity. Novel mediators of metastasis have been identified by comparing clusters of tumor cells, captured by using a negCTC-iChip, with single CTCs from breast cancer patients: single cell-resolution RNA sequencing demonstrated the significantly divergent expression of only a few genes, including plakoglobin. Plakoglobin functions as an intercellular tether that confers added metastatic potential to tumor cells in the circulation. Knockdown of plakoglobin inhibits CTC-cluster generation and lung metastases, while both abundance of CTC-clusters and high plakoglobin levels denote adverse outcome in breast cancer patients, pointing to CTC-clusters as critical mediators of cancer metastasis.

7Recent data support the hypothesis that CTC profile changes occur during tumor cell dissemination, in order to enter into the vasculature, survive in circulation, extravasate, and generate secondary tumors. Thus, in response to signaling proteins released by stromal cells, tumor cells undergo reversible phenotypic change, described as epithelial-to-mesenchymal transition (EMT). EMT is also required in tissue repair; consistently, the metastatic process is at least in part similar to tissue regeneration. EMT results in a reduced expression of epithelial markers, and an increased plasticity and capacity for migration and invasion, as well as a resistance to anoikis and apoptosis, and senescence, which are hallmarks required for CTC survival and dissemination.

8Studies of the effects of EMT in CTCs have suggested that mesenchymal transformation may facilitate the initial steps of the metastatic cascade, but may reduce their competence to initiate overt metastases. Disseminated cancer cells may at least partially revert to an epithelial phenotype to promote adhesion and proliferation in distal sites, through mesenchymal–epithelial transition. Cancer-associated EMT may therefore not represent a complete interconversion of phenotypes, or an irreversible commitment to a mesenchymal state. Indeed, the existence of a metastable or partial EMT phenotype, with both epithelial and mesenchymal features, adapts to the concept of a highly plastic stem-like state, reversible and responsive to the microenvironment.

9Several groups investigated this phenotype in breast cancer. At least one of a selected pool of EMT markers (Twist 1, Akt2, or PI3Kalpha) were observed both in CTC-positive and CTC-negative cancer patients. Overexpression of EMT-initiating transcription factors (TF), including Twist, Snail1, Slug, Zeb1, and FoxC2, was detected in CTCs, even if in a minority of patients. The EMT markers fibronectin and/or vimentin were expressed more frequently in cytokeratins (CK)-negative than in CK-positive samples. Of note, patients receiving neoadjuvant therapy were more likely to present a higher expression of EMT markers, predicting worse prognosis more accurately than epithelial markers in patients with CK-negative CTCs; thus, CTCs, subjected to EMT, might be invasive cells resistant to neo-adjuvant therapy.

10Several lines of evidence have coherently confirmed the notion that the metastatic potential of a tumor is due to a low number of a minor subpopulation of cancer cells (~1%)—termed cancer stem cells (CSCs)—in primary tumor tissue, that exhibit two defining abilities, to self-renew and to efficiently regenerate the phenotypic heterogeneity of a parental tumor, and which are responsible for initiating overt metastases. Studies of neoplastic tissues have provided evidence of stem-like cells within tumors, of breast (CD44+CD24−/low), pancreatic (CD44+CD24+ESA+), brain (CD133+), head and neck (CD44+), and colon (CD133+ and CD44+ESAhigh) cancer. CSCs have been implicated in both initiating primary tumors and in the seeding and establishment of metastases. Recent studies have pointed to a crucial link between EMT and the acquisition of stem cell properties. In addition to conferring migratory and invasive potential, the induction of EMT in immortalized and transformed human mammary epithelial cells significantly enhanced their self-renewal, tumor-initiating capabilities, and led to the expression of stem-cell markers, associated with breast CSCs, suggesting that passage through EMT is an alternative and/or additional step in tumorigenesis. Indeed, the induction of EMT has been associated with expression of the CD44+CD24−/low antigenic profile, which, in breast tumors, defines a subgroup of enriched cancer cells with stem-like properties, as well as mesenchymal traits and cancer stem cell properties, including self-renewal capabilities and resistance to therapies. These cells can self-renew in vitro to generate both CD44highCD24low cells and CD44lowCD24high cells, providing further evidence of stemness properties. Aberrations in signaling networks and transcription factors implicated in EMT have been documented in breast cancer progression, and numerous conserved genes in the signature of resident stem cells are considered EMT-associated genes. Consistently, the gene expression profile of CD44+CD24−/low breast cancer cells more closely resembles that of CD44+CD24−/low cells from normal breast tissue. This undifferentiated phenotype of CSCs is evident in basal-like breast tumors, which exhibit EMT features accounting for their clinical seriousness. Further, the gene expression signatures of stem cells from normal tissues and of the highly aggressive metaplastic and claudin-low histological breast tumor subtypes share strong similarities with gene expression profiles of cells upon the induction of EMT and cancer stem cells.

11Both in vitro and in vivo experiments have suggested that long-term maintenance of the EMT/stem cell state depends on continuous EMT-inducing signals. EMT is orchestrated by integrated networks of signal pathways and EMT-related TFs, including TGFβ, NOTCH, and Wnt. Exposure of tumor cells to both cytokines TGFβ and TNFα induced EMT and generated cells with a stable breast cancer stem cell phenotype, enriched in CD24−low/CD44+ cells, with a shift to the claudin low molecular subtype. Targeting proteins required for induction of the claudin-low subtype could be a therapeutic strategy for improving the survival of breast tumor patients, regardless of the primary subtype. TGFβ represents a paradigm of duality in cancer. Of note, the initial tumor suppression effect due to TGFβ is coupled to EMT; however, although the bulk of tumor cells die, EMT-permissive cells can survive from the TGFβ-imposed bottleneck. Moreover, tumorigenicity linked to EMT is a trait that must be selected during carcinoma progression. In accordance, EMT is not always connected with effective metastatic dissemination.

12Several TFs and miRNAs have been implicated in linking EMT to the acquisition of stem cell properties. EMT-inducing TFs have been confirmed to confer malignant traits in neoplastic cells, and the acquisition of stem cell properties of self-renewing and tumor-initiating, thus generating highly aggressive tumors with an EMT phenotype and stem cell features, and providing evidence for the emergence of cells with combined EMT/CSC phenotypes. The discovery of miRNAs has added an additional level of complexity in EMT, metastasis, and stemness. Downmodulation of the miR-200 family has been extensively documented to occur during EMT, and has been described both in invasive pancreatic tumors, inducing the generation of migrating CSCs, and in normal and breast cancer stem cells, reinforcing the links between normal stem cells and CSCs. Additionally, expression levels of let-7, which is poorly expressed in mesenchymal tumors and targets HMGA2, encoding a DNA-binding protein implicated in mesenchymal cell differentiation, are reduced in breast CSCs, but increased during differentiation.

13EMT enables the detachment of tumor cells from a primary site, but it can also be induced after CTCs enter the vasculature. Only recently, different subsets of CTCs have been identified as mesenchymal. However, exclusively mesenchymal CTCs seem to be unable to establish metastases. The current hypothesis is that tumor cells with an intermediate phenotype between epithelial and mesenchymal present the highest plasticity to adapt to secondary sites, and constitute cancer stem cells. Consistently, CTCs expressing EPCAM, CD44, CD47, and MET, once injected in immunodeficient mice, generated metastases with an EPCAMlow/METhigh/CD47high/CD44high phenotype, thus containing functional metastasis-initiating cells. Therefore, the detection of such CTCs is of the utmost importance.

14The use of EpCAM in current CTC detection methods, thus effectively pre-selecting for epithelial CTCs, is a matter of debate, and high expression has been confirmed in CSC from breast, colorectal, and pancreatic tumors. The metastatic potential of EpCAM-positive CTCs has been extensively confirmed in xenograft models, providing the proof of principle that CTCs captured by the CellSearch system are capable of initiating metastasis, at least in murine models. On the contrary, the CellSearch method did not recognize aggressive breast cancer cells, and it suffered from a relatively low sensitivity: only a fraction of patients with metastatic cancer scored positive for any CTCs, with a median yield of approximately 1 CTC per milliliter and typically low purity.

15Although the role of EMT in CTCs is still controversial, EMT signatures have been observed in at least a subset of tumor cells with stemness properties. However, studies in cancer models suggest that EMT could be dispensable for establishing metastasis, despite the fact that it contributes to the aggressiveness by increasing chemoresistance. Some investigators reported that CSCs markers were enriched in CTCs with mesenchymal features and had prognostic relevance, in breast tumors at different stages (stage I–III). These CSC phenotypes need to be further evaluated, also in patients with early stage cancer, and prospectively related to the occurrence of overt metastases. Furthermore, CSCs might develop after the initial diagnosis of a tumor, so serial postoperative blood monitoring might be required to identify CSCs.

16A selection of the more aggressive cells generates clinical evidence of metastasis. Coherently, a significant genetic heterogeneity in early disseminated tumor cells (DTC) in early breast cancer contrasts with homogeneity in tumor cells from metastases. Subsets of CTCs/DTCs demonstrate a stem cell-like phenotype, which confer stemness and pleiotropic roles in cell adhesion, migration, and homing. Only a small subset of CTCs, which migrate into vasculature, termed circulating cancer stem cells (CCSCs), with putative stem cell progenitor phenotypes, present high invasiveness and effectively generate metastasis. Blood-monitoring studies in tumor patients have shown that some CTCs can survive chemotherapy, as postulated for cancer stem cells. CCSCs have been identified in several types of cancer, including metastatic breast cancer, lung carcinoma, hepatocellular carcinoma, colorectal cancer, and gastric cancer. Specific biomarkers associated with stemness detected in CTCs include BMI1 for lung CCSCs, CD90+CXCR4+ for hepatocellular CCSCs, and CD44 variant exon 9 (CD44v9) for colorectal CCSCs.

17Stem cell and epithelial-mesenchymal transition markers are frequently overexpressed in CTCs of metastatic breast cancer. Recently, in a breast tumor, a stem cell-like population, defined by the presence of CD44 and absence of CD24, has been identified, which might disseminate into the circulation and escape therapy, and presents an expression profile associated with metastatic relapse. It has been reported that the modulation of HER2 signaling can increase the cancer stem cell pool and might be required for its maintenance; in fact, a strong link between HER2- like-tumors and stem cells has been observed. Preliminary data identified a breast cancer stem cell-like phenotype in CTCs, with increased resistance to chemotherapy and decreased proliferation in circulation. ALDH1, a marker of normal and neoplastic breast stem cells, has been confirmed to be overexpressed in 70% of CTCs, and associated with therapy failure. At least one or more of the EMT markers Twist, Akt2, and PI3Kalpha, and ALDH1 were detectable in a great proportion of CTCs, thus identifying the highly tumorigenic subset of EMT-associated breast CSCs. This subset of CTCs is indicative of therapy-resistance and inferior prognosis in metastatic tumor patients, with clinical relevance. Of note, ALDH1 and EMT markers were measured even before the detection of CTCs in circulation, as assessed through the positivity in RT-PCR to at least one of the transcripts HER2, MUC1, and EpCAM. Consistently, the stem cell-like phenotypes CD44+CD24−/low and ALDH1highCD24−/low have been identified in 35.2% and 17.7% of CTCs, respectively, and the subset of CD44+CD24−/low CTCs among ALDH1-highly positive CTCs present an even higher tumorigenic potential. Finally, although ALDH1-positive cells represented only 5% of cells in tumors expressing ALDH1, ALDH1 staining positivity was associated with a high histological grade and poor clinical outcome.

18According to the cancer stem cell model, tumor progression and drug resistance should depend on putative cancer stem cells, and their presence should correlate with a worse outcome, despite data being discordant. Effective elimination of these cells during medical intervention is difficult, due to their self-renewal potency and resistance to chemotherapeutics. Indeed, patients with ALDH1-positive CTCs more frequently did not respond to chemotherapy. Recently, in 86% of metastatic breast cancer patients, CTCs have been demonstrated to express one or more multidrug resistance-related proteins (MRP), and patients with MRP-positive CTCs had a significantly shorter time to progression. Further studies are required to correlate the presence of CD44+CD24− or ALDH1highCD24−/low CTCs with clinical course and disease progression.

19CSCs of different metastatic sites might have different phenotypes. In HCC, stem-like CTCs were considered as seeds of metastasis and recurrence. A significant positive correlation was measured between CD133+ CTCs and serum levels of Annexin A3 (ANXA3), which is highly accurate in diagnosing early-stage HCC, and in monitoring the therapeutic response. Prognosis was worse for ANXA3-positive patients with detectable CCSCs, suggesting that serum ANXA3 could stimulate and maintain the stem cell-like traits of CD133+ CTCs to promote tumor recurrence and metastasis. In breast tumor patients, both CK+CD44+CD24−/low CTCs, or CTCs expressing CD44, MET, and CD47, might represent CSCs. Moreover, CTCs from patients with metastatic breast cancer, competent in generating brain metastases, were EPCAM− and expressed HER2, EGFR, NOTCH1, and HPSE.

20EpCAM-negative CTCs may efficiently avoid organ arrest due to the presence of stem cell and quiescence properties. The molecular switch to differentiate quiescence in malignant CTCs depends on interactions with the tumor microenvironment. Previous studies have established that the loss of the urokinase plasminogen activator receptor (uPAR) and integrin β1 (intβ1) reduces proliferative signals, with a shift from an invasive or metastatic to a dormant tumor state. In order to identify CTC subsets with properties related to breast cancer dormancy, CTC subgroups were selected for EpCAM and CD45 negativity and positivity for the CD44+CD24− stem cell signature, and the combinatorial expression of uPAR and int β1. A high expression of embryonic stem-cell genes was confirmed in the uPAR+intβ1+ CTC subset, compared to uPAR−intβ1− CTCs. Accordingly, in vitro assays confirmed the metastatic competency of uPARintβ1 CTCs, consistent with previous notions. The characterization of the uPARintβ1 CTC subset can be useful to decipher cellular and molecular mechanisms of organ-homing CTCs and cancer dormancy.

21The liquid biopsy of CTC has received great attention, due to obvious clinical implications for personalized medicine. Clinical applications of CTC detection have been extensively described in recent reviews. Strong evidence for CTCs as prognostic markers of metastatic relapse and progression has been documented. The clinical utility of CTCs enumeration for treatment decisions is currently being evaluated in interventional studies. CTC detection has been proven to predict prognosis also in early-stage tumor patients, without clinical signs of overt metastases. Detection of CTC at the time of diagnosis or after surgery for a primary tumor can identify patients at risk of recurrent or metastatic disease. Moreover, the patterns of drug susceptibility and responses to therapeutic regimens in individual patients, as tumors acquire new mutations over time, could be noninvasively monitored through the characterization of CTCs, as confirmed in a recent proof-of-concept study of drug sensitivity testing in ER–positive breast cancer CTCs. Most agents administered to prevent relapse are designed to target growing cancer cells rather than quiescent DTCs that predominate in metastatic latency. Research on these cells should help in designing novel therapeutic molecules for targeting and monitoring minimal residual disease, and to assess response to therapy.

22Unfortunately, the detection of CTCs is hampered by their still uncertain biology. Profiling of CTCs in serial liquid biopsies holds great potential to significantly improve cancer therapy by: (i) determining patient prognosis, (ii) monitoring tumor recurrence and therapeutic responses, (iii) identifying new therapeutic targets in order to prevent metastasis, (iv) elucidating drug resistance mechanisms, and (v) improving current understanding of tumor progression and metastasis.

23At present, insufficient data on the clinical relevance of stem-like CTCs is still present in the literature, mainly due to technical hurdles in detecting the very rare stem-like tumor cells among CTCs. Nonetheless, as novel non epithelial profiles of CTCs have started to emerge, in this manuscript we underline their relevance to patient management. It is possible that, as for the mesenchymal CTCs, stem-like CTCs will also be proven to bear clinical implications, as suggested from the reported early studies.

24The development of single cell analysis technologies has demonstrated a great heterogeneity in the genetic make-up of CTCs, compared to primary tumor and metastatic cells. Exclusive genetic mutations of CTCs have been identified, while additional mutations in CTCs were also detected in both subclones of primary tumor and metastatic tissues, as assessed in diverse tumors. Discordances have been reported in the expression of cancer-specific antigens between CTCs and primary tumors. CTCs represent an effective alternative to invasive tumor biopsies, thus constituting the actual targets of adjuvant therapies, and point out the emergence of resistant clones and patterns of early disease recurrence and therapy responses. CTCs in women initially diagnosed with advanced ER-positive/HER2-negative breast cancer acquired an HER2-positive subpopulation after multiple courses of therapy for recurrent metastatic tumors. Although they have comparable tumor-initiating potential and cell plasticity and expression profiles of ALDH1, these coexisting discrete chemotherapy-sensitive HER2+NOTCH1− and NOTCH inhibitor-sensitive HER2−NOTCH1+ CTCs interconvert, thus a dual treatment may be required for an effective response. Also, markers relevant to the breast cancer brain-metastasis-selected CTC profile have been reported, and the uPARintβ1 CTC subset may prospectively identify patients at high risk. In breast cancer, the oncogene PIK3CA is closely associated with the reactivation of multipotency, and a complete response is more difficult in HER2-positive tumors mutated in PIK3CA after neoadjuvant anthracycline taxane-based chemotherapy plus anti-HER2 treatment, even with a dual anti-HER2 treatment. A multimarker panel based on single CTC analysis, which defines resistant tumor cells in ovarian cancer by detecting epithelial, EMT, and stemness-associated transcripts, has recently been designed, confirming that the number of CTCs is negatively correlated with patient overall survival. In metastatic castration-resistant prostate cancer (CRPC) patients, a set of EMT-related genes (PTPRN2, ALDH1, ESR2, and WNT5A) were upregulated in CTCs; upon targeted therapy two distinct CTC subgroups were induced, which differ in both AR genotype and genomic CNV profiles. Thus, serial analyses of CTCs represent a reliable test for the molecular evolution of tumors and may be relevant upon effective targeted cancer therapy that induces drug resistance.

25As for EMT in CTCs, the presence of mesenchymal markers can accurately predict clinical outcomes and a poor prognosis. In HCC metastasis, Twist and vimentin expression can be applied as diagnostic and prognostic biomarkers; of note, the expression levels of Twist, vimentin, and E-cadherin were closely associated with the positivity rates of CTCs in peripheral blood, which indicated that EMT promotes the dissemination of primary HCC cells. In breast cancer patients, several differentially expressed genes, involved in EMT, were identified in CTC, including UPA, IGFR1, VEGFR1, and CD44, and their expression was correlated with tumor grade and metastasis. Activation of the urokinase plasminogen system, which has been associated with poor prognosis, and IGFR1 both contribute to recurrence via the inhibition of anoikis. VEGFR1, a positive regulator of angiogenesis, participates in tumor progression through the induction of EMT. Expression of CD44 has been detected in breast primary tumors and significantly increased in CTC-enriched samples, in accordance with a previous report. In pancreatic tumor expression of both epithelial and mesenchymal markers, LGFBP5 and ALDH1 have been detected in CTCs.

26The clinical relevance of mutation analysis of CTCs in assessing prognosis, stratifying patients, defining precise therapies, and monitoring therapeutic efficacy has been confirmed. In breast cancer, the analysis of single CTC with a diagnostic intention identified preexisting cells resistant to ERBB2-targeted therapies. In colorectal cancer, several mutations that induce resistance to anti-EGFR therapy have been reported; CTC monitoring has revealed that the emergence of KRAS mutant clones is a secondary mechanism of drug resistance. The acquired S492R EGFR mutation prevents cetuximab binding, and confers resistance. In non–small-cell lung cancer (NSCLC), the diagnosing of EML4-ALK rearrangement, a potent oncogenic driver, is a critical issue for identifying patients with advanced tumors that eventually respond to crizotinib. A recent study provides the first proof-of-concept that CTCs can be used for highly sensitive and specific diagnostic testing of ALK rearrangement. A unique pattern of ALK rearrangement has been consistently identified in CTCs, despite inter-tumoral heterogeneity. Additionally, results support a role for ALK in generating mesenchymal CTCs, suggesting that CTCs with a unique ALK rearrangement and a mesenchymal phenotype may result from the clonal selection of tumor cells that have acquired the potential to metastasize, such as cancer stem cells.

27As mentioned above, recent advances in the identification of stem-like signatures in primary tumors and CTC may help to improve the accuracy of diagnosis and predict therapeutic responses and overall survival of patients. Molecular profiling of CSCs has revealed their major implications for clinics. Consistently, the phenotypic features of CTCs, including CCSCs, have been analyzed to assess tumor stage, risk of cancer progression and metastases, and disease relapse, and their potential in predicting the overall outcome of patients with locally advanced and metastatic breast, prostate, ovarian, pancreatic, lung, and colorectal cancers has been confirmed. Several biomarkers of stemness have been assayed in CTCs (CD133, CD44, ALDH, and ABC multidrug transporters). In breast cancer, the expression of Nanog, Oct3/4, Sox-2, nestin, and CD34 on CTCs is directly related to disease progression. Recently, a very small cohort of metastatic breast cancer patients was evaluated after palliative chemo-, antibody—or hormonal therapy; most patients received anthracyclines and taxanes in an adjuvant or metastatic setting before the study’s start, and all patients with HER2+ tumors but one received trastuzumab in the metastatic setting. In this cohort, the expression in CTCs of at least one EMT marker and ALDH1 was detected in 62% and 44%, respectively, of patient non responders to therapy, suggesting their potential use as predictive biomarkers. The hypothesis of plasticity and stemness properties in CSCs has been addressed in endometrial carcinomas (EC), in association with a micro-RNA signature of EMT of down-modulation of the miR-200 family. The presence of endometrial CTC, profiled for stem-cell markers ALDH and CD44, was related to recurrent disease. CTCs detected in patients with advanced lung cancer include cells co-expressing cytokeratin and CD133. In a cohort of patients with primary and metastatic melanoma, the melanoma antigen melan-A and the stem cell–like marker ABCB5 were expressed in CTCs in 45% and 49% of patients with recurrence after treatment initiation, respectively.

28A major clinical relevance of CTC early detection is the identification of high-risk subjects that need additional systemic therapies after primary tumor surgery, to prevent metastasis. Currently, the selection of patients is based on the statistical risk of recurrences. Evidence for the presence of CSC in high-risk EC has been reported, and a molecular CTC-phenotype has been identified, associated with plasticity and stemness features. Previously, a six-gene panel has been assessed in CTCs in EC, with a potential in terms of diagnosis/prognosis and follow-up, and identification of new therapeutic strategies to limit metastatic dissemination. Several genes related to plasticity were significantly expressed in CTC (ETV5, NOTCH1, Snai1, TGFB1, Zeb1, Zeb2). Of note, ETV5 up-modulation promoted metastasis in in vivo models and recapitulated the plasticity phenotype of high-risk patients. The sequential assessment of CTC levels during treatment could provide information at early stages about therapeutic efficacy.

29Moreover, CSCs, generated through EMT, present intrinsic resistance to conventional chemotherapies. Current chemotherapies target the rapidly proliferating cells of the tumor bulk, with expansion of the CSC pool and/or selection of resistant and highly metastatic competent CSCs. Accordingly, in breast tumors, residual cancer cells exhibited expression profiles of cells with combined tumor-initiating and mesenchymal/claudin-low features. Gene signatures of the EMT/CSC phenotype may disclose novel molecular targets for the therapeutic resistance of breast cancers. In a seminal study, salinomycin, a potassium ionophore, has been identified with selective toxicity against CSCs generated by EMT, reducing the CD44highCD24−/low fraction and the ability of pretreated CSCs to initiate tumors and lung metastases in immunocompromised mice. Further, metformin selectively killed breast CSCs in vitro and in vivo, effectively reducing tumors and preventing relapse in mouse models. Overall, cells with the EMT/CSC phenotype may uncover new therapeutic targets for tumors, irrespective of their subtype.

30The rarity of CTCs in blood specimens (as few as one cell per 109 haematologic cells) requires highly specialized technologies for detection and capture, but their low sensitivity has limited the clinical utility of CTC-based diagnostics. A lot of reviews have recapitulated and pointed out a conceptual framework of CTC assays, discussing limits of technologies, in light of the recent data (see Table 2 for a list of the most relevant technologies for CTC detection and characterization). Epithelial markers, expressed on epithelial tumors but absent on leukocytes, have been frequently used. While clinical studies using the CellSearch device have confirmed the prognostic potential of CTC enumeration, only a fraction of patients with metastatic cancer score positive for CTCs. Although first defined as nucleated cells expressing EpCAM, and/or cytokeratin, and lacking the leukocyte antigen CD45, new data suggest even greater heterogeneity in CTCs. Considering the phenotypic heterogeneity and plasticity of CTCs, CTCs with no/low levels of EpCAM and a more mesenchymal phenotype, such as CSCs, might be missed by EpCAM-based capturing techniques.

31To effectively capture all diverse CTCs, additional alternative phenotypes need to be detected, and an automated and high-precision device integrating CTC capturing and sequencing is needed. Due to recent advances in the field of microfluidic bioengineering, novel methods have been introduced to improve the accuracy and sensitivity of CTC isolation, with efficient purification of viable CTCs from unprocessed whole blood, in both epithelial and non-epithelial cancers. Once CTCs have been sorted in solution, clinically standardized analyses can be applied, as well as single-cell RNA profiling. To avoid the loss of CTCs, normal haematopoietic cells have been depleted by targeting CD45. However, not all CD45– cells are tumor cells, and a CK+/CD45+ cell phenotype has been reported in metastatic cancer patients. Examples of technologies used to isolate CTCs through both enrichment and depletion include MACS, a size-dictated immunocapture chip, and the CTC-iChip. The CTC-chip enables a high yield of capture and purity, although it is technically difficult and not yet standardized for high throughput applications. The MagSweeper has been used for the genetic profiling of single-cell CTCs, including transcriptional profiling, the detection of mutations, whole exome sequencing, and analyses of stem cells in several cancers.

32High-throughput microscopy and imaging flow cytometry IFC can be combined with cutting-edge CTC isolation technology. One such approach currently under investigation relies on the concept of an inertial microfluidic device, and staining of CTCs with anti-CD45, anti-cytokeratin, anti-E-cadherin, and anti-vimentin. Once clinically validated, this technology, adapted for EMT and stem cell markers, would provide a technological solution for implementation in clinical diagnostics.

33A novel enhanced CTC enrichment strategy was established, which captures heterogeneous EpCAM-depleted CTC subsets by targeting cell surface proteins (CD44, CD47, c-Met, reported to be expressed in CTCs with metastasis-initiating properties, Trop2, CD49f, CK8, ADAM8, TEM8, CD146), as well as extracellular matrix components (laminin, collagen I, HA). In 20 out of 29 EpCAM-depleted fractions from metastatic breast cancer patients, additional EpCAMneg CTCs were identified applying Trop2, CD49f, c-Met, CK8, and/or HA magnetic enrichment. Of note, EpCAMneg cells captured by a CD44 antibody were confirmed to have a malignant nature. Further characterization of identified EpCAMpos/neg and dual-positive subsets will assess CTCs tumor origin. To prove the clinical relevance, a study identifying EpCAMlow/neg CTCs with specific genomic profiles should be designed.

34A selective and sensitive detection method for the isolation of both CTCs and CCSCs is required for advancement in precision medicine. Fluorescence-activated cell sorting has been used for CCSC isolation by comparing a CSC marker (CD133 or CD44) with leucocyte CD45. Overall, only a limited success in CCSC detection has been reported by using immune-affinity-based methods with multiple fluorescent probes. To address this, a novel nanoparticle-mediated Raman imaging method has been described for the detection and analysis of heterogeneous CTC subtypes, with a high isolation efficiency, which profiles CCSCs in accordance with surface marker expression. The selective detection and accurate analysis of tumorigenic properties of CCSCs and CTCs were demonstrated. Five different RANs, as multifunctional probes designed for the detection, isolation, and further analysis of circulating cells through Raman imaging, were designed that identify different CTC subtypes and CCSC’s stemness, by targeting CD133. Transplanted CCSCs could proliferate (CD133+ and CD44+CD24−/low), as well as differentiate, into subtype-specific breast cancer cells in metastatic cancer modeling, including HER2 positive and basal-B. By detecting CCSC, cancer metastasis and relapse would be predicted, and the heterogeneous tumorigenicity of CTCs interrogated, to facilitate the study of chemotherapy effects.

35Stem-cell-like CTCs have also been detected through PCR-based approaches. In CRC patients, the detection of CTCs positive for CEA/CK19/CK20/CD133 expression represented a significant prognostic factor for poorer DFS and OS in Dukes stage C patients, and in stage B patients with unfavorable pathological features. In addition, CEA/CK/CD133+ CTC could be a marker of treatment-resistant stem cells. In a recent study, which aimed to identify stem cell markers for CTSC detection, the differential expression of several stem cell markers was observed between CRC patients’ tumor and matched normal tissue, including OLFM4 and LGR5. Furthermore, plastin 3 (PLS3) is expressed in EMT-induced CTC in CRC, thus identifying tumor cells with a down-modulated expression of epithelial markers; the detection of such cells was an independent prognostic factor for progression-free (PFS) and overall (OS) survival, also in the Dukes B subset. These findings suggest that OLFM4, LGR5, and PLS3 may be useful for the detection of CTSCs.

36The presence of CTCs at different stages of prostate cancer has been assessed, using the ProstateCancerDetect kit and the AdnaTest EMT-2/StemCell designed panel. Positive results were mainly seen in patients with metastatic disease; EMT-like gene expression was exclusively detected in metastatic tumors, while ALDH1 was even increased in localized tumors. Thus, in advanced prostate cancer, CTCs mainly express transcripts associated with progression and metastatic disease. Even so, the test’s prognostic and predictive implications remain to be determined. The expression of several markers has been analyzed by immunocytochemistry in CTCs from patients with CRPC. Most of the CTCs co-expressed CD133 with epithelial and mesenchymal markers. A quantitative PCR method has also been used to detect EMT (Twist1 and vimentin) and stem cell genes (ABCG2, CD133, PSCA) in peripheral blood from patients with metastatic prostate cancer; expression of stem cell-related genes indicates poor prognosis, whereas EMT-related expression does not. Recently, we described a method based on a qRT-PCR assay for the identification of novel CTC markers, irrespective of currently used markers and without pre-selection or enrichment, by using patient-derived xenografts (PDXs) models; the genes identified using this method in PDX are now also being tested for the study of CTC heterogeneity in human patients.

37In conclusion, the reports we discussed here suggest a compelling need to address the definition of common criteria describing critical subtypes of CTCs. During the progression of cancer, EMT can confer properties of stem cell state and acquisition of malignant traits by more differentiated neoplastic cells. Of clinical relevance, patients with CTCs presenting a stem cell-like or EMT phenotype respond less to chemotherapies and show shorter progression-free survival. Detection of this type of CTC could be a powerful diagnostic tool for patient stratification and for the early determination of failure of a therapeutic intervention. As proof, gene signatures and stem cell–like markers expressed in CTCs have been identified for non-invasive prognosis. The molecular targeting of CSCs and CTCs represents a potential strategy to counteract cancer progression, and to overcome treatment resistance and relapse. Additionally, the identification of CSC and CTC targets might contribute to addressing the chemo-resistance of circulating CSCs, a major problem in cancer biology and therapeutics.