Circulating Tumor Cell Detection Technologies and Clinical Utility: Challenges and Opportunities

1The advent of new diagnostic and treatment modalities have improved the 5-year relative survival rate for all types of cancers combined; the survival rate increased substantially from 39% to 70% among white patients and from 27% up to 63% among black patients. It is widely accepted that the capture, enumeration and identification of circulating tumor cells (CTCs) hold significant promise for early cancer screening, diagnosis and prognosis. These cells originate from primary tumors and disseminate to distant sites via the blood (Figure 1). The rate of tumor shedding has been estimated to be around 3.2 million cells per gram of tumor per day. The majority of these cells get rapidly cleared, but those that survive bloodstream shear stress and evade recognition by immune cells are speculated to have the highest metastatic potential. Hence, the peripheral blood offers a rich reservoir of cancer-derived materials beneficial for cancer diagnosis and monitoring. Moreover, there is a rich body of literature that supports the diagnostic and prognostic clinical value of CTC analysis in metastatic breast, prostate, lung, and colorectal cancers. In this work, we review the clinical readiness of various attempts to develop, and commercialize, CTC detection technologies, and discuss challenges facing their clinical utility for cancer diagnosis, prognosis, and management.

2Several CTC detection platforms have emerged over the past decade, each exploiting a distinctive characteristic of CTCs for sensitive selection and capture. Each technology differs in the biophysical or bimolecular trait leveraged for CTC capture, enrichment, and downstream cellular and molecular characterization, but all of them aim to enumerate CTC and draw clinically relevant conclusions regarding the prevalence of CTC for cancer management. The focus of these methodologies is the detection of CTC clinically, rapidly, and with high sensitivity, selectivity and specificity, while remaining minimally invasive. The differences between CTC and normal blood cells in gene/protein expression, morphology, volume, and biophysical properties had led to the establishment and commercialization of several CTC detection and enumeration devices during the past decade. These commercialized technologies can be categorized based on method of CTC identification as label dependent (affinity-based) or label independent. Then, each category is subdivided into different classes based on its functional detection approach (Figure 2).

3The most widely used approach for CTC detection and isolation is immune-based detection, whereby antibodies are used to selectively bind cell surface antigens. Tumor cells express different cell surface markers than blood cells and therefore can be separated from the circulatory cells. More specifically, CTC expresses epithelial markers like epithelial cell adhesion molecules (EpCAM) and Cytokeratin (CK) to variable degrees, but not CD45, a differential marker for while blood cells (WBCs). The antibodies can be conjugated to magnetic nanoparticles or immobilized on the walls of microfluidic chips for CTC capture through the positive selection of CTC or the negative depletion of white blood cells. This approach is also termed affinity-based isolation and capture of CTC, and it is the most commonly used technique in Clinical Laboratory Improvement Amendments (CLIA)-certified molecular diagnostics laboratories. A full review of the commercially available CTC label-dependent technologies is summarized in Table 1.

4One of the leading platforms utilizing label-dependent technology is the CellSearch systems (Veridex), which employ EpCAM-coated ferrofluid nanoparticles for the selection of EpCAM-positive CTCs followed by confirmation with immunostaining for the high expression of CK 8, 18 and the absence of CD45 expression. More than 10 CellSearch kits are available commercially and can identify clinically relevant populations of CTC. One kit received Food and Drug Administration (FDA) approval in 2004, the circulating epithelial cell in vitro diagnostics (IVD) kit. CellSearch has been widely used in monitoring patients with metastatic breast, colorectal, lung or prostate cancer. The first clinical trial with CellSearch was conducted by Cristofanilli and colleagues, where they measured levels of CTCs in metastatic breast cancer; their trial indicated that CTCs could be used as a marker for overall survival (OS) and progression-free survival (PFS). Ever since, CellSearch technology has proven to be sensitive and reproducible in a large number of studies.

5Important correlations between CTC count and cancer relapse have already been observed using the CellSearch system, but the technology has some limitations. Firstly, it relies on the expression of EpCAM only and ignores other potential biomarkers. EpCAM has a dynamic expression among CTCs of different origins and at different cancer stages; it becomes highly downregulated when CTCs disseminate from primary sites and undergo epithelial-to-mesenchymal transmission (EMT) to acquire a more aggressive phenotype and seed for metastasis. Studies have shown that EpCAM’s presence has been correlated with localized cancer; however, upon metastasis, its expression, along with that of CK, decreases amid the appearance of mesenchymal markers. Secondly, cell isolation through the CellSearch system is followed by cell fixation for stabilization, which prevents further viable cell characterization such as CTC culture, CTC 3D organoid culture or CTC-derived xenografts. Another major pitfall in the CellSearch system is the low purity of captured cells which ranges between 60–70%, the captured CTCs are usually contaminated with blood cells or normal circulating epithelial cells (CEC). In addition, the CellSearch system has a low sensitivity for CTC detection (one cell per 1 mL of a blood sample).

6AdnaTest® (AdnaGen AG, Langenhagen, Germany) is another commercialized positive selection platform that relies on immunomagnetic beads coated with a cocktail of antibodies for the enhanced capture and enrichment of CTCs in breast, prostate, ovarian and colon cancer. The positively enriched cells are then tested by multiplex RT-PCR for various gene panels (e.g., MUC-1, HER-2 and GA733-2) based on tumor type for CTC validation, an improved design that combines two methodological approaches for enhanced sensitivity and better heterogeneity characterization of CTCs. AdnaTest® offers physicians the opportunity to analyze the clinical relevance of CTCs by testing the gene expression of specific tumor markers in the captured cells and comparing this with their primary and metastatic tumor counterparts, thus possessing a diagnostic and prognostic value on one hand and also allowing the manipulation of treatment regimens. In a clinical study to assess the prognostic validity of CTC in metastatic castration-resistant prostate cancer, AdnaTest detected CTCs in 62% (34/47) of patient samples compared to only 45% with CellSearch (23/47). This difference could be attributed to the fact that AdnaTest identified the CTCs based on the presence of KLK3, PSMA and EGFR transcripts compared to CellSearch, which depends only on EpCAM for capture. In this study, the overall sensitivity of the CellSearch results was considered unfavorable. Although Adnatest combines two approaches in its design, in some cases, CellSearch has superiority because of the technical difficulties accompanied with AdnaTest blood sample processing and preservation. For example, in a comparative metastatic breast cancer clinical study, the CellSearch system was able to detect CTCs in 50% (116/221) of patients compared to AdnaTest, which detected CTCs in 40% of the patients only (88/221). The CTCs detected by AdnaTest had no association with OS nor with PFS in contrast to prognostic outcomes in CellSearch.

7In addition to CellSearch and AdnaTest, various technologies depend on immunomagnetic beads for CTC enrichment, such as Magnetic-activated Cell Sorting (MACS) systems (Miltenyi Biotec, Bergisch Gladbach, Germany), and MagSweeper (Illumina, San Diego, CA, USA) (reviewed in Table 1). MACS offers both positive enrichment and negative enrichment for the high-efficiency isolation of CTCs. On the other hand, MagSweeper processes larger volumes of blood (9 mL/h) and can detect 1–3 CTCs per 1 mL of unprocessed blood. However, there are unavoidable problems with this approach, the reliance on EpCAM and CK presence and the variation in their expression among the different types of cancer and the different stages is a major drawback for these modalities. In addition, the use of expensive antibodies leads to high detection costs and the internalization of micro- and nanomagnetic affects the viability of captured cells. Moreover, post-capture, CTCs cannot be further analyzed in real-time and live-cell conditions, since the cells are fixed or lysed during the assay process.

8Microfluidic chips are capable of integrating conventional biological assays at the microscale that can lead to CTC detection, isolation, and even cell culture. Hence, these platforms show a major potential for processing CTCs post-capture and isolation. The precise microfabrication used to build these platforms offer controlled geometries, fluid flow and surface functionalization, leading to directed and regulated cell contact with the chip walls. The relatively high surface areas coated with antibodies provide many sites for CTC capture. The captured cells are fixed on the micro-spot, while remaining blood components are carried away by the flow; this allows nonspecifically bound cells to be washed out, thus ensuring the high enrichment of CTC against leukocytes. The most commonly used antibody is EpCAM, but several chips employ a cocktail of antibodies specific to a particular cancer or use anti-CD45 and/or anti-CD66 antibodies for negative depletion (retaining of leukocytes and eluting CTCs). Today, several devices have been proposed for this functionality, including the geometrically enhanced mixing (GEM) chip, geometrically enhanced differential immunocapture (GEDI) chip, LiquidBiopsy® (Cynvenio Biosystems, Inc., Thousand Oaks, CA, USA), and OncoCEE (cell enrichment and extraction) (Biocept, San Diego, CA, USA). Compared to immunomagnetic bead-based platforms, these devices have higher capture specificity and selectivity, and thus can be used for diagnosis and not only prognosis. A novel surface-based microfluidic capture system for CTC selection with high throughput, high recovery, high purity and full automation is the modular CTC sinusoidal microsystem (BioFluidica, LLC., San Diego, CA, USA). The system is composed of three modules; the first is the selection module, which has 50–320 parallel sinusoidal channels coated with antibodies (anti-EpCAM, and/or anti-seprase/FAP alpha) for CTC capture, the second is the impedance sensor module for label-free counting, and the third is the collection module for collecting captured cells that are stained and imaged consequently. This microsystem can detect up to 50 CTCs/mL with >85% purity. The microfluidic and specifically ligand-based capture approach we discussed allows for the release of cells post-capture for phenotypic processing and molecular analysis. However, some limitations still exist, including limited volume and slow rate processing as well as shear forces that might affect cell viability and cell–ligand attachment if not sufficiently low.

9This is a new generation of CTC antigen-based capture platforms that combine immunomagnetic beads with the advantages of microfluidics to address the shortcomings of single modality platforms. The CTCs are usually isolated from whole blood via immunomagnetic beads, then passed through microfluidic platforms to isolate the magnetically labeled cells according to size or inertial focusing for higher sensitivity capture (Figure 2a). A promising example is the CTC-iChip device, which houses three microfluidic components inline for CTC capture. Firstly, the cells are magnetically labeled in whole blood, then the cells are passed sequentially in three microfluidic technologies within a single automated system. The first component is deterministic lateral displacement, which removes nucleated cells from whole blood samples based on cell size; the second component is inertial focusing to line the nucleated cells, and the third component is magnetophoresis to magnetically deflect magnetically labeled cells from unlabeled cells into a separate channel. This platform has a dual format; it either uses positive selection (CTC-iChippos) using EpCAM-labeled magnetic beads or negative selection (CTC-iChipneg) using CD44- and CD66b- labeled magnetic beads. In comparison to CellSearch, CTC-iChippos, where CTCs are captured with EpCAM beads and stained for CK, displayed higher sensitivity to the capture of CTC in patients with low CTC count (<30 CTCs/7.5 mL) and achieved a >3.5-log purification (a background of 1500 WBCs/mL of the blood sample). In a head-to-head comparison with CellSearch in clinical samples with <30 CTCs per 7.5 mL, the number of isolated cells by CTC-iChippos was significantly higher in 22 out of 36 samples (p < 0.001). The isolated cells were then molecularly characterized by RT-PCR, four positive samples were correctly identified and correlated to patients with an EML4–ALK oncogenic fusion protein at their primary tumor, an approach that could not be attained with CellSearch. It is worth mentioning that the CTC-iChip is compatible with standard CTC analysis post-capture protocols and is being developed for diagnosis purposes. With dual modality approaches, the enrichment efficiency is usually higher (>99%) with a higher purity of isolated cells. However, this platform is still immature and not widely investigated. Another promising example is the Ephesia system, which was developed initially to capture leukemic B-cells and then modified to capture CTCs. The sample is made to flow through a diamond-like chip into the capture columns zone. Supermagnetic beads coated with EpCAM antibodies self-assemble into a periodic array under a high magnetic field that creates a dense sieve, which, in turn, captures the passing EpCAM positive cells. This technology avoids the need for costly and complicated microfabrication and can be modified to have different sizes and geometries with varying rates of flow. Ephesia is currently not commercialized but has been shown to be promising for diagnosis, especially with its high capture specificity. In a study comparing Ephesia cell capture technology with the reference CellSearch, CTCs were detected in clinical samples taken from metastatic breast cancer (4/5) or metastatic prostate cancer (6/8) patients. The Ephesia method showed a higher quantity of captured CTC compared to CellSearch in 10 out of the 13 samples. Each capture modality, single or dual, is highlighted in Table 2 with a comparative analysis of their advantages and disadvantages.

10Given that CTCs express EpCAM and CK to varying levels, with some displaying the complete downregulation of these proteins, alternative strategies have been developed and tested to isolate and enumerate CTCs based on their biophysical properties. These platforms discriminate CTCs from other cells based on physical characteristics such as size, density, deformability, and electrical properties. A cellular analysis of CTCs has revealed that they have a greater nucleus-to-cytoplasm ratio, larger sizes (>10 µm compared to <5 µm), a different nuclear morphology, and exhibit different electrical properties than normal cells. In the following section, these differences are highlighted as the detection principle and are summarized in Table 3.

11Microfiltration enrichment methods process whole blood through a range of microscale constrictions to capture target cells based on their size or a combination of size and deformability. The recovery efficiency is reduced due to the buildup of filtration resistance resulting from the frictional drag on the blood sample as it passes through the filter, a main limitation of this strategy.

12The most significant physical difference between CTCs and WBCs is size, where CTCs are larger on average. Several platforms aim at sieving CTCs from a blood sample and have been shown to be more selective and efficient than the CellSearch system. These platforms rely on microfiltration, which involves a single membrane with varying pore sizes between 6 and 9 µm and is used to separate CTCs and filter out smaller blood cells. A novel filter-based size exclusion technology called ISET (isolation by size of tumor cells), developed by RareCell Diagnostics, Paris, France, was capable of isolating CTCs independent of their expression of any particular marker. Using this technique, CTCs were detected in patients with hepatocellular carcinoma, breast carcinoma, and melanoma. In a new study, ISET could detect single CTCs in 80% of 40 evaluated patients with early-stage resectable non-small-cell lung cancer (NSCLC), compared to only 23% of the patients using CellSearch. On the other hand, circulating tumor microemboli (CTM, clusters of ≥3 CTCs) were captured in 80% of patients using ISET but were undetectable by CellSearch. Of the captured single CTCs, 62% were positive for the proliferation marker Ki67, whereas cells within CTMs tested negative. Bobek V. et al. used a new size-based separation modality, the MetaCell® technology, for the enrichment and cultivation of CTCs in vitro. Using this technology, CTCs were detected in 66.7% of patients, with comparable frequencies in patients with operable and inoperable tumors (60% vs. 77.8%). Comparable CTC fractions were observed among patients with metastatic and nonmetastatic tumors (66.7% vs. 66.7%). The CTCs were then cultured in vitro for further downstream applications, thus confirming their viability. Although simple and easy to automate, the filtration system is prone to pore clogging, and requires large volumes of blood. Moreover, the purity of the collected samples has been an issue in some devices.

13Microfluidic chips have also been developed for CTC size-based detection and termed “three-dimensional microfiltration”, where 3D geometries are constructed and designed to allow the separation of CTCs from background blood components. The Parsortix system is an example of such an approach: it has a stair-like architecture that decreases gradually in width down to 4.5 µm to aid in the capture of larger cells (i.e., CTCs) and provide the necessary physical support above and below the captured cells to prevent morphological changes. CTCs larger than the channel width become trapped in the gaps, while smaller cells pass through. Its design maximizes the length of separation and allows reverse flow for the subsequent release of captured CTCs for downstream interrogation and analysis.

14Density gradient centrifugation is a typical method for segregating whole blood into its constituents based on differences in sedimentation coefficients. As whole blood is dropped in the liquid gradient while being subjected to centrifugation, cells are distributed along the gradient depending on their density. Erythrocytes or polymorphonuclear leukocytes with lower cellular density are precipitated at the bottom, whereas the heavier mononuclear leukocytes and CTCs remain at the top. Several platforms have been developed for preclinical and clinical applications, the most common being OncoQuick® (Greiner Bio-One International GmbH, Frickenhausen, Germany) and Percoll, Ficoll-Hypaque® (GE Healthcare Life Sciences, Uppsala, Sweden), mostly used in biomedical laboratories to recover peripheral blood mononuclear cells. Despite its long history of use in laboratory environments, there are some drawbacks linked to this technique, mainly the possible loss of CTCs that move either to the plasma region or to the bottom of the density gradient due to the formation of aggregates. It is worth noting that this cell loss could be due to the cytotoxicity of the density medium. Interestingly, the Percoll density (GE Healthcare Life Sciences) gradient medium has some advantages over Ficoll, which include reduced toxicity as well as a wider density gradient range. Another platform that uses gradient centrifugation to separate cells is the Cyttel system, which combines anti-CD45 immunohistochemistry and fluorescence in situ hybridization followed by gradient centrifugation and slide smearing to identify CTCs. Tong et al. performed a comparative clinical study between Cyttel method, and the CellSearch platform on 127 patients. The CellSearch system revealed that only 30–50% of lung cancer patients had more than one CTC in 7.5 mL of blood, and just 20–30% of patients had more than two CTCs in 7.5 mL of blood. Using the Cyttel method, CTCs were present in 87% of the patients; among them, 63% had a CTC count of ≥3 cells/3.2 mL, 37% had a CTC count of ≥5 cells/3.2 mL, and 15.2% had a CTC count of ≥8 cells/3.2 mL.

15Inertial focusing uses the effects of fluid inertia in microchannels of a certain shape to align microparticles and cells at high flow rates. When randomly dispersed particles, such as cells in blood, flow through a channel with a particle Reynolds number of one or greater, they are subjected to two counteracting inertial lift forces: a force that directs particles toward the channel walls, and another that repels the particles toward the channel centerline. In square or rectangular channels, combining these forces leads to the migration of particles to two to four dynamic equilibrium positions located between the channel centerline and the wall. Following focusing, cells are collected in a smaller volume and significantly concentrated in a size-dependent fashion. Sollier E. et al. combined this phenomenon with laminar micro-vortices to trap CTCs until flushed out by perfusion at flow rate; the tailored approach is called Vortex. Their design makes use of multiple expansion-contraction reservoirs placed in series and parallel, which generate multiple vortices when a laminar flow of a sample occurs at a high rate. CTCs are then simply released, after a wash step that removes any remaining small particles, by lowering the flow rate. Using this system, Sollier E. et al. were able to extract and enumerate CTCs from four patients with breast cancer and eight patients with lung cancer. From 7.5 mL of blood, 25–51 and 23–317 CTCs were detected from breast and lung cancer patients, respectively.

16Another modality that makes use of inertial focusing is the ClearCell FX, developed by Clearbrdige Biomedics. ClearCell FX is a spiral microfluidic device that combines both inertial focusing with the secondary Dean’s flow resulting from curved channels to trap CTCs from a blood sample; this allows for the proper positioning of CTCs within the channel. This modality can process 7.5 mL of blood in less than 10 min but requires red blood cells lysis prior to sample processing. Khoo et al. tested this system on patients with metastatic breast cancer or NSCLC. CTCs were detected in 100% of the patients, with a varied range of CTCs isolated for breast cancer samples (12–1275 CTCs/mL) (Median: 55 CTCs/mL) and NSCLC samples (10–1535 CTCs/mL) (Median: 82 CTCs/mL), respectively. Inertial focusing allows the recovery of viable cells for downstream analysis and does not require complex high-resolution imaging techniques nor the use of expensive antibodies. The emerging evolution of this modality will allow the creation of low-cost CTC detection chips for better cancer management.

17In addition to size-dependent isolation, other physical traits observed in CTCs are exploited to distinguish them from leukocytes. Two innovative approaches to cell separation are dielectrophoresis and direct imaging, which both depend on cell composition, morphology, and phenotypes. These two platforms are discussed in the following subsections.

18Imaging-based detection makes use of specific fluorescent tags to identify and count CTCs in blood. Several imaging-based CTC detection technologies have been developed and tested and each is unique in its sample preparation, detection algorithm, and fluorophores used. Somlo G. et al. developed a novel fiber optic array scanning technology (FASTcell™) to detect CTCs in patients with locally advanced/inflammatory breast cancer (LABC/IBC), metastatic breast cancer (MBC) and non-small cell lung cancer. The principle of operation of the FASTcell™ technology is based on an array of optical fibers that form a wide collection aperture to allow a wider field of view; this enables the rapid high-fidelity localization of CTCs identified by conventional markers such as CK, DAPI and CD45 without the need for enrichment. The FASTcell™ system can scan a sample of blood on a glass side at a rapid rate of 25 million cells/min, comprising the image resolution and subsequent confirmation of potential detected CTCs. Clinical studies using FASTcell™ have successfully identified CTCs in 62% of LABC/IBC patients, in 82% of MBC patients and 42% of non-small cell lung cancer patients.

19Another novel imaging approach that combines both flow cytometry with fluorescence imaging for high throughput analysis has been developed recently. The technology is referred to as ImageStream, and its detection of CTC depends on the expression of EpCAM, CK, AFP, glypican-3 and DNA-PK together with an analysis of size, morphology and DNA content. In a clinical study of patients with hepatocellular carcinoma (HCC), between one and 1642 CTCs were detected in the blood samples of HCC patients (45/69) compared to zero CTCs in the controls (0/31). In a comparative study between CellSearch and ImageStream, the number of counted CTCs from a reference value did not differ significantly between both methods, however ImageStream had a lower level of precision when fewer CTCs were analyzed. ImageStream, if developed properly and tested extensively, can serve as a platform for CTC enumeration for early diagnosis.

20Dielectrophoresis (DEP) is a liquid biopsy separation method that relies on particles with different polarization that move differently under a nonuniform electric field. Microchips that use the DEP technique have been developed for isolating and capturing CTC via multiple integrated electrodes, generating a nonuniform alternating electric field. The Apostream™ system by ApoCell was the first commercially available DEP field flow fractionation where CTCs and leukocytes are separated based on the differences of their conductivities. Its methodology requires an initial enrichment step. The recovery rate recorded is over 70%, the viability is higher than 97% and the processing time is 1 h for a 10 mL blood sample; however, the purity obtained is less than 1%. DEPArray, on the other hand, is the second commercialized DEP strategy to separate CTCs in a blood sample using controllable electrodes. This platform is designed for single CTC capture, allowing downstream gene analysis and sequencing. In spite of the advantages demonstrated by DEP-based detection and capture methods, there are some limitations. Among these are the low sample volumes and the varying dielectric characteristics of cells due to ion leakage; this limits the isolation time. In addition, the electric resistance of the running medium used must be low, which is not achievable for all samples studied, especially those from diabetic patients.

21As in label-based enrichment technology, label-free enrichment methodologies have their own advantages and disadvantages that are presented in Table 4.

22In 2015, a search for “circulating tumor cells” in the “ClinicalTrials.gov” website revealed 296 studies involved in CTC detection and capture in patients with metastatic disease. Currently, in 2020, the same search engine showed 867 studies on CTCs registered in the U.S. National Library of Medicine website, 271 of which are complete, leaving 596 studies in progress. Of these completed studies, only 60 have published their results with promising outcomes. The fact that the number of clinical studies involving CTCs has doubled in less than four years highlights the clinical and commercial interest in their potential. On the other hand, the low number of promising results point to serious challenges facing the clinical utility of CTC detection technology. The latter includes demonstrating high accuracy in capturing viable CTCs discussed in the previous sections and predicting clinical outcomes of therapeutic regimens which we will discuss in this section.

23Among the clinical uses of CTC detection is cancer prognosis, which is significant for clinical decision making as prognostic estimation is useful for assessing the risks and benefits of the proposed treatment. Huge efforts have been made to understand the clinical utility of CTCs to predict prognosis and guide therapeutic decisions. We selected seventeen studies to demonstrate and confirm the prognostic, and sometimes the diagnostic, implications of CTCs in various types of cancer using different CTC detection technologies including CellSearch. In these studies, like in most cases, patients who had a higher count of CTCs (unfavorable count) have had a worse prognosis, measured primarily by PFS and OS survival estimates. The results of these trials shed light on the real possibilities of CTC counting and gave deeper insights on the potential of using CTCs as liquid biopsies (Table 5, Figure 2b).

24The pretreatment staging of patients with cancer and risk stratification based on clinical and histopathological findings greatly improved prognostication and treatment allocation for many cancer subtypes; however, there remains much room for improvement. Evidence show that the implementation of CTC detection and quantification can improve the accuracy of patient preoperative staging. In a recent study conducted on 100 preoperative patients with esophageal cancer, CTC detection using the CellSearch system was correlated with clinicopathological parameters for measuring the prognostic outcome. CTCs were detected in 18% of all eligible patients. Compared to the histopathologic stage (invasion, tumor grade and histological subtype), CTC positive patients had inferior OS (Hazard Ratio: 3.128, p < 0.001) and relapse-free survival (RFS) (HR: 5.063, p < 0.001) compared to CTC negative patients. Even in patients with non-metastatic tumors and lymph node invasion, CTC presence indicated worse OS and RFS. A multivariate analysis identified CTCs as strong independent prognostic indicators of tumor recurrence (HR: 5.063, p < 0.001). The outcome of this study suggests the clinical relevance of CTCs as preoperative prognostic and staging parameters in esophageal cancer. In a similar study with CellSearch, CTC detection in preoperative patients with colorectal cancer (stages I–IV) has been used as a prognostic marker with a threshold of at least one CTC per 7.5 mL. CTC detection was more frequent in patients with metastatic compared to those with non-metastatic disease. Patients with non-metastatic CTC positive disease have a worse OS (38.4 months) compared to CTC-negative patients (49.8 months) in the same cohort (p < 0.001). Importantly, multivariate analysis revealed that CTC detection was the strongest prognostic predictor independent of other clinicopathological parameters; in fact, there was no association between primary tumor characteristics or clinicopathological parameters (age, sex, disease site, T stage, N stage, resection margins and distribution of metastasis) and CTC detection in non-metastatic patients. These studies provide evidence-based utility of CTCs as predictive biomarkers which can be further used to stratify patients’ risks within the different stages of the disease. However, the use of CellSearch only investigates a subpopulation of CTCs (EpCAM and CK positive), ignoring a broader spectrum of cells without enumeration or analysis. In addition, such studies must be validated in further cohort forms in multi-institutional trials with longer follow-up times in order to evidentially conclude whether CTC detection should be included in prognostic measures and treatment guidelines.

25The prognostic significance of CTC is also true for patients undergoing a new line of treatment. Using the CellSearch platform, predictive outcome measures of CTC count and chromogranin A (CgA) were investigated in 138 patients with metastatic neuroendocrine neoplasms receiving a new line of treatment (somatostatin analogues, chemotherapy, PRRT, and TAE). Of all patients, 51% had received previous anticancer therapy, 41% were receiving long-term SST (somatostatin analogue), and 60% of patients tested positive for CTC (at least one CTC detected). Fifteen weeks after starting the new line of therapy, patients with zero CTC count had the highest OS (49.1 months), patients with a CTC count from one to eight had lower OS (21.5 months, 95% CI: 1.20–5.16), and patients with a CTC count above eight had the worst OS (13.3 months, 95% CI: 2.70–9.74). The changes in CTC count significantly correlated with the response to treatment and OS, suggesting their potential use as surrogate markers to direct clinical decision-making. On the other hand, changes in CgA were not significantly associated with survival.

26In another study, a pooled analysis of 1944 patients with metastatic breast cancer from 20 different studies at 17 European centers validated the prognostic clinical utility of CTCs. Before using a new line of treatment, 47% of patients tested positive for CTCs using the CellSearch system (threshold ≥ 5CTCs/7.5 mL). Patients with elevated CTCs at baseline had an inferior OS (HR: 2.27, p < 0.0001) and PFS (HR: 1.92, p < 0.0001) compared to patients with a CTC count lower than five CTCs/7.5 mL. After the new line of treatment, any increase in CTC count correlated with shortened OS and PFS. When the CTC count was added to the full clinicopathologic predictive models, the prognosis accuracy was improved according to using the likelihood ratio (LR) χ2 statistical analysis. It is worth mentioning that serum tumor markers (CEA and CA15-3) did not show any significant prognostic value, even when added to the clinicopathologic predictive model. However, the surrogacy of CTC count as a method to manage treatment was not assessable in this study because the data was not collected from randomized therapeutic trials. In other words, assessing switching treatments based on CTC count was not evident in assessing the response to treatment. Similar patterns were obtained in other studies listed in Table 5, all of which confirm that elevated levels of CTC are clinically important markers that reflect an unfavorable tumor response to treatment and worse patient outcomes.

27Beyond diagnosis and guiding prognosis when enumerated, CTCs have been used as pharmacodynamic biomarkers to assess inherited or acquired resistance in response to specific treatment regimens. The molecular characterization of CTC has influenced the choice of therapeutic options in some cases. It also has improved our understanding of the underlying mechanisms of cancer metastasis and helped in discovering new targets for therapeutic manipulation. For example, in a phase I clinical trial, CTCs were assessed based on their count and the expression of insulin-like growth factor-1 receptor (IGF-1R) to determine therapeutic approaches. The CellSearch system was used to count CTCs expressing IGF-1R in patients treated with monoclonal antibodies against IGF-1R, either alone or in combination with docetaxel. Out of 26 patients, 23 had positive IGF-1R CTCs and had responded better to the combinatorial treatment compared to the remaining three patients whose CTCs were negative for IGF-1R. Such a study suggests the potential use of CTCs as predictive markers for the choice of administered chemotherapy. Another example is a cross-sectional cohort study carried by Scher and colleagues on 193 patients with mCRPC; the EPIC platform was used to correlate the expression and localization of the nuclear androgen-receptor splice variant 7 (AR-V7) in CTCs in response to taxanes and androgen receptor-signaling inhibitor (ARSI). Overall, patients with nuclear AR-V7 positive CTCs had a poor prostate-specific antigen response, and a short PFS and OS when treated with ARSI, while those treated with taxanes had a much lower risk of death compared to the first group.

28Several clinical studies have reported the use of CTC with circulating tumor DNA (ctDNA), tumor-derived DNA released into the blood via apoptosis or the necrosis of shedding cells from primary and metastatic lesions, as a complementary biomarker for treatment assessment. In a phase II clinical trial of erlotinib and pertuzumab in patients with advanced NSCLC, a decrease in CTC count upon treatment was correlated with longer PFS. In addition, patients with EGFR mutations showed a substantial reduction in CTC count throughout treatment. The mutational analysis of EGFR showed that ctDNA had higher sensitivity in detecting mutations compared to CTC and, upon treatment, a decrease in mutational load suggested a partial response to treatment. This study, and another mentioned in Table 5, suggest the potential role of CTC combined with ctDNA as biomarkers for the early indication of treatment response.

29In the past two decades, our improved understanding of the biology of CTCs and their role in cancer metastasis has opened the door to a myriad of technologies aimed at exploring the clinical potential of CTCs as biomarkers for cancer. CellSearch remains the only successful platform to obtain FDA approval, while many are still in preclinical and clinical trial stages. Antigen-free approaches are showing a lot of potential for clinical success as they overcome the heterogeneous expression of expressed membrane proteins in CTCs. Affinity-based detection platforms are primarily designed to detect, and count CTCs based on the expression of epithelial markers, mainly EpCAM and CK, but such markers are proven to be downregulated upon undergoing EMT. Such inadequacy necessitates either finding a universal selection epithelial and mesenchymal cell surface marker to prevent cross-reactivity with other types of cells in the blood or adopting physical, mechanical and electrical properties as selection markers for isolating CTCs from their surroundings. In addition, any selection approach also may need to distinguish single cells and clusters (few currently do, see Table 1). The presence of the latter has been linked with a poor prognosis.

30Owing to CTC heterogeneity and the different approaches used to assess the sensitivity and specificity of each detection platform, as highlighted in Table 1 and Table 3, it is often hard to comparatively evaluate the output of these different platforms in diagnosis, prognosis, and cancer screening. The interpretation of such metrics is quite challenging because of some major limiting factors such as the volume of blood used, the size of the cancer patient population, the used mathematical model for evaluation and, most importantly, the reference for comparison (i.e., CellSearch, blood sample, other microdevices, etc.). Several clinical studies have highlighted the prospects for monitoring cancer patients, while others have used CTC to dissect its biological metastatic potential. One of the exciting clinical applications for CTC technology is correlating CTC counts with clinical outcome measures such as OS and PFS. Most published studies showed that CTC enumeration was the strongest prognostic biomarker for patients’ survival rates, even with multivariate analysis.

31On the other hand, despite the few promising results obtained with CTC enumeration and molecular characterization for prognostic predictions and treatment response, implementing CTC count as a method to manage treatment in cancer patients is not attainable. In a cohort clinical trial (SWOG S0500) in patients with metastatic breast cancer and no CTC decrease in response to the first-line chemotherapy, switching to second-line chemotherapy did not affect CTC count, nor did it improve OS. This means that although an elevated CTC count is correlated with ineffective treatment response, changing the treatment while considering CTC count only might not be beneficial. The clinical utility of CTCs needs to be validated through randomized clinical interventions in which therapy decisions are based on CTC analysis and established endpoints. Additionally, comparing the outcome of the different CTC clinical studies is challenging due to the use of different patient populations, different CTC detection technologies, different definitions of threshold positivity, different statistical methodologies, and different clinical endpoints used for outcome reporting. To date, standardized methods for clinical CTC isolation and identification are still lacking, contributing to controversial outcome measures in some cases. In addition, the heterogeneity and rarity of CTCs, along with the relatively small cohort of patients tested, might influence the findings and interpretations of the various prognostic and diagnostic outcome measures being tested. With ambiguities in the clinical relevance of CTC, hardly any CTC detection technology is going to be introduced and implemented in routine clinical practices any time soon. Thus, a shift from just counting CTC into a more detailed molecular and functional characterization of CTC will certainly provide a comprehensive picture of the malignancy using only patient blood. In addition, complementing CTC analysis with other liquid biopsy biomarkers such as circulating tumor DNA (ctDNA) can enable inclusive results to better understand the tumor status of a patient. The fact that ctDNAphylogenetic analysis can depict early stages of cancer evolution and elucidate the therapeutic resistance of tumors to chemotherapy through genetic mutations suggests its clinical importance. Such combination studies using two liquid biopsies are promising, but the performance characteristics have been modest and require further investigation.

32Despite all the advances in CTC detection technologies and their diverse capture and enrichment systems, many significant challenges are yet to be met, particularly those with respect to analytical and clinical sensitivities. Adopting such tools into routine clinical practice will demand laborious studies into their analytical validity, clinical validity and clinical utility. These tools, coupled with bioinformatics tools and annotated databases, will provide evidence as to whether detected genomic aberrations in blood may aid in predicting the most suitable cancer therapy on a personalized level.