Circulating tumour cells: a broad perspective

Circulating tumour cells (CTCs) have recently been identified as valuable biomarkers for diagnostic and prognostic evaluations, as well for monitoring therapeutic responses to treatments. CTCs are rare cells which may be present as one CTC surrounded by approximately 1 million white blood cells and 1 billion red blood cells per millilitre of peripheral blood. Despite the various challenges in CTC detection, considerable progress in detection methods have been documented in recent times, particularly for methodologies incorporating nanomaterial-based platforms and/or integrated microfluidics. Herein, we summarize the importance of CTCs as biological markers for tumour detection, highlight their mechanism of cellular invasion and discuss the various challenges associated with CTC research, including vulnerability, heterogeneity, phenotypicity and size differences. In addition, we describe nanomaterial agents used for electrochemistry and surface plasmon resonance applications, which have recently been used to selectively capture cancer cells and amplify signals for CTC detection. The intrinsic properties of nanomaterials have also recently been exploited to achieve photothermal destruction of cancer cells. This review describes recent advancements and future perspectives in the CTC field.


Introduction
Circulating tumour cells (CTCs) constitute an exceedingly small fraction of cells relative to a background of 1 million white blood cells (WBCs) and 1 billion red blood cells (RBCs) per millilitre of peripheral blood [1]. CTCs and circulating tumour microemboli (CTM) are the intermediate stages in metastasis [2,3]. The goal of cancer invasion is territorial dominance and eventual cell death [4]. The key cellular process involves (i) intravasation, (ii) CTCs extravasation to bone marrow or other organs where the cells are disseminated and metastasized at the local site and (iii) CTM may also undergo intravascular proliferation to bone marrow or other organs before the tumour cells are disseminated and metastasized at the local site [3]. The spontaneity of these processes and the identification of CTCs/CTM and leucocytes in the blood have been the subject of considerable interest over the years. Therefore, understanding the role of CTC in a diseased cancer patient is paramount for long-term survival. Furthermore, approaches toward CTC capture and detection, cell type enumeration and the subsequent application of relevant therapeutic measures are essential for the management of cancer patients. Despite the extremely low concentration of CTCs in body fluids, they can potentially provide information regarding diagnosis, prognosis and follow-up of therapeutic responses to treatments [5]. CTCs are therefore considered valuable biomarkers in targeted molecular therapies for cancer patients and are also candidates for determining CTC phenotypes in preclinical models [5][6][7]. The extremely low concentration of CTCs in peripheral blood samples makes isolation, enumeration and detection daunting. Moreover, due to a purification step that is required for their characterization, captured CTCs may become damaged during the isolation process.
However, CTCs are challenging to accurately enumerate and detect even by the current standard method, CellSearch ® (Veridex), which remains the only test technology approved by the Food and Drug Administration (FDA) [1]. As illustrated in figure 1, the CellSearch method preselects based on epithelial-based adhesion molecule (EpCAM) expression [8]. Additionally, there are still no clear guidelines for optimal CTC enumeration for clinical decisionmaking in treating primary and metastatic breast cancers [9]. Furthermore, CellSearch is based on the enrichment of CTCs using the EpCAM or negative depletion of leucocytes [1,10]. Although this technology has recently been used to demonstrate the clinical relevance of CTCs as an independent and predictive marker for early breast cancer [11], prognostic evaluation remains widely conjectural for many reported clinical cases including: (i) non-metastatic colorectal cancer patients [12] and (ii) tumour cell recurrence in patients with localized prostate cancer [13]. However, in small cell lung cancer (SCLC) and non-small-cell lung cancer (NSCLC) patients, Cell-Search remains the validated approach for the prognostic evaluation of CTCs [14][15][16]. For instance, CTCs are highly detectable in patients with stage IV NSCLC [14] and SCLS [15]. Therefore, the CellSearch technology must be reviewed carefully on a case by case basis. Importantly, CellSearch can only detect CTCs that express EpCAM (EpCAM + CTC) and not EpCAM with low or no CTC (EpCAM − CTC). Therefore, when Wit et al. combined platforms, they were able to capture (EpCAM − CTC) using a filtration and fluorescent staining protocol and EpCAM + CTC was captured with CellSearch technology. Thus, their combined platforms increased CTC detection in the blood samples of 27 metastatic lung cancer patients to 41% as opposed to 15% detected by CellSearch only, which is indicative of a good outcome in the study [10,17]. However, CTC with EpCAM − affinity has not been confirmed from large pool studies and molecular characterization of this marker from EpCAM + remain undifferentiated. The Xu et al. [18] study compared CellSearch to their optimized Parsortix system-an example of an epitope-independent method. The group recovered significantly more CK + CTC than the CellSearch method as well as capturing CTC clusters from 7.5 ml of 10 prostate cancer patient samples. In another study, Kulasinghe et al. [19] compared CellSearch with two epitope-independent approaches in advanced stage head and neck cancer (HNC) patients. The results obtained for single CTCs isolation with different technologies included: (18.6%) Cell-Search, (46.4%) ScreenCell and (64%) by RosetteSep™ including detection of CTC clusters. The role of EpCAM negative CTCs is not fully understood-whether they are prognostic has not been investigated [20].
In this review, we emphasize nanomaterial-based assays for nanoscale sensing by electrochemical methods and surface plasmon techniques. Other in-depth areas covered include a molecular understanding of the mode of action of cancer invasion in tumour biology and CTC characterization. We also discuss some critical challenges facing existing technologies. In addition, we emphasize the importance of surface plasmon resonance (SPR) and surface-enhanced Raman scattering (SERS) as complementary tools for biolectrochemical detection processes. Finally, we cover the potential of nanotheranostics to provide future outpatient precision therapeutics for cancer.

Origin of metastatic invasion
Due to the complexity of metastasis, the molecular nature of cancer is not fully understood. However, in the classical seed-and-soil hypothesis established by Stephen Paget in 1889 [21], cancer is described as the cross-talk between selected cancer cells (the seed) and the specific organ environment (the soil). It has been extensively reviewed by .
(1) Paget's hypothesis states that carcinomas are biologically heterogeneous, containing subpopulations of cells having different regulatory pathways and invasive attributes, and undergoing metastatic processes. (2) Metastasis occurs through complex regulatory pathways known as invasion-metastasis  royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 cascades, which include the adaptation of invasion features such embolization (covering CTCs with platelets); CTC survival in circulation; arrest in distant capillary beds; and extravasation into, and multiplication within, the organ parenchyma (functional organ site only). (3) Metastasis depends on multiple interactions (cross-talk) within metastatic cells that may eventually be controlled by tumour cells. (4) The intravasation mechanism may be further divided into single CTCs and CTM. The single CTCs may undergo extravasation via bone marrow or other organs where the cells are disseminated and metastasized at the local site. Another regulatory pathway is that CTM may be developed via intravascular proliferation to bone marrow or other organs before the tumour cells are disseminated and metastasized at the local site [3]. The inherent morphological attributes of CTC clusters include [33]: (i) they are rarer than CTCs, (ii) these clusters are formed by oligoclonal tumour cell groupings, whose origin may be related to biclonal gammopathies-where two or more distinct proteins are synthesized [34], (iii) CTC clusters have 23-50× higher metastatic potential than single CTCs. Additionally, the importance of circulating clusters has recently been highlighted in breast cancer and human glioblastoma models [35,36]. Plakoglobin presence has been identified as the probable cause of CTC cluster formation in breast cancer cell but the relationship of these two variants in patients remains evaluated [33]. Further readings on CTC clusters have been covered by Hong et al. [37].
The most recent updates on the seed-and-soil hypothesis reviewed by Akhtar et al. [21] and Massagué et al. [38], described the invasion journey, summarized in figure 2. The tumour cells that break away from the primary tumour lose their epithelial properties and acquire mesenchymal-like properties during the intravasation stage [39]. The epithelial-mesenchymal transition (EMT) [39] is mediated by cadherin molecule switching (calcium-dependent cell adhesion), involving the downregulation of Ecadherin and upregulation of N-cadherin. The modulation of E-cadherin and N-cadherin levels are indicative of metastatic breast cancer [4,40]. Furthermore, during EMT, upregulation of vimentin, integrin-ανβ6 and metalloproteinase also occurs. Upregulation of N-cadherin is indicative of invasive metastatic breast cancer types [41,42]. In metastasis, integrin-ανβ6 binds several ligands from the extracellular matrix, thus maintaining upregulation activity and increasing migration survival and inactivation of apoptosis. Integrin-ανβ6 is usually overexpressed in breast cancer metastatic lesions during the regulation process, and it plays a key role in tumour growth, invasion, early angiogenesis activity and metastasis [43,44]. Mesenchymal cells then undergo structural changes that enable the invasion of the surrounding extracellular matrix, stroma and platelets. CTCs interaction with platelets helps to provide protection against destructive physical forces from the surrounding environment through the activation of thrombosis and fibrin deposition. This step is known as the embolization stage. Because of the various immune defensive systems and the mechanical stresses of blood flow, only the few CTCs that escape the primary tumour invade the surrounding host tissues or organs. The invading cells at this stage are known as disseminated tumour cells (DTCs), which is the transition from mesenchymal to epithelial cells. Less than 0.01% of CTCs survive to the next stage of tumour latency [45]. Finally, after the extravasation stage, DTCs settle into a dormancy lasting for several weeks to several decades; these cells may eventually undergo micrometastasis and produce metastatic lesions. Some characteristics of cancer cells and major areas of clinical research advancement are highlighted in table 1.

Challenges associated with CTC-based technologies
The challenges associated with CTCs have been highlighted in numerous review articles [46][47][48] [175] royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 reviewed is the mode of CTC release after capture to enable culture expansion of CTC phenotypes, highlighted by the Kelly group [49] as one of the key areas for development of next-generation materials in this area. There is also a paradigm shift from a two-dimensional (2D) tissue-culture model to three-or four-dimesional (3D or 4D) microenvironments for subsequent CTC expansion [50,51]. Interestingly, the advent of imprintable materials sensitive to environmental changes such as solvents, heat, mechanical stress or other external stimuli may pave the way for a more advanced shape than 3D, which is a shape-shift environment in a 4D system. With a 4D imprintable material, it will be possible for materials to exist in two stable forms (exhibiting zero degree at each stage of the freedom) in response to environmental changes such as solvents, heat, mechanical stress or other external stimuli that effect change. More importantly, these microenvironments enable dynamic, spatio-temporal cellular behaviours to be modulated under physiological conditions and provide a utility model for drug screening [52][53][54].

Biological challenges
The current challenges hindering the efficient use of CTCs in biological environments include CTC vulnerability and viability, and the inability to analyse viable CTCs through intracellular staining. Additionally, using the current FDA-approved standard, CellSearch, CTCs may remain undetected because of a lack of detectable biological markers for recognizing other important non-EpCAM markers. Thirdly, CTCs have phenotypic heterogeneity, making enumeration difficult after isolation. CTCs also have a mixture of phenotypes, among which only cells positively expressing the antigen EpCAM are counted using CellSearch, thus ignoring EpCAM − CTC. Furthermore, captured CTCs may also be damaged during release, hindering downstream characterization. A recent review by Green et al. [49] highlights the mechanism of CTC release, which may occur via chemical, enzymatic, self-assembly-based and mechanosensitive modes or via a thermal release mechanism (figure 3). Pertinent here is that next-generation materials that can effectively capture and release CTCs have recently emerged and these materials enable culturing and expansion of CTCs, post capture, and provide an excellent platform for understanding CTC subpopulations and heterogeneity [46][47][48][49].

Technical challenges
The capability of any device to enrich CTCs present at typically 1-100 cells ml −1 of whole blood is considered a significant performance evaluation matrix [55][56][57]. However, the performance of CTC multifarious approaches may affect data quality, specifically around capture efficiency evaluations, CTC heterogeneity, size variability, non-specific cellular labelling initiated during capture, analysis of unaccounted CTCs, misidentification of damaged CTCs to include them into the overall population figures, and finally limitations regarding blood sample volume and assay robustness in CTC capture [58,59]. Therefore, statistical considerations for any CTC device should be elucidated. For example, in one study [59], a cell-based assay calculation for flow cytometry of blood volume for high precision system was determined using the following equation: where r represents the number of events that meet the required criterion. CV is the coefficient variation of a known positive control. From equation (1.1), if the desired CV is 10%, 100 events are required, and the true value would be between 80 and 120 events, which the authors' state is close to the limit of detection for most laboratory cytometry tests. The same estimated value of the CV approximately 10% has been described as the cut-off value for improving the accuracy of the assay [58]. The statistical risk in this probability sampling is in setting a threshold cut-off for the number of CTCs captured as a direct way to evaluate a cancer patient's prognoses as either good or bad. Another consideration that may have unwanted statistical and technical consequences is CTC damage caused by high flow-induced shear [60]. In one study, the microfluidic environment was modulated to generate a broad range of  royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 haemodynamic shear stresses [60]. Briefly, high shear stress causes CTC stress, thus inducing necrosis (unregulated cell death), preventing CTC attachment and inducing 90% necrosis within the first 4 h of circulation and apoptosis within 16-24 h. Additionally, it was observed that high shear stresses are effective at killing other types of epithelial-based cancer cells as well as drug-resistant breast cancer cells.
Phenotypic heterogeneity of CTCs is also a challenge (due to the complexities of the statistical analysis). Some researchers have used sophisticated algorithms, including the naive Bayesian classifier which is based on a probabilistic generative mixture model [61], to identify individual cells in a sample. Although complex statistical procedures are involved, the model is based on (i) identifying images to be processed and implementing an algorithm, (ii) subdividing the images processed, and (iii) re-identifying and grouping the images. The advantage with this platform is that the Bayes theorem is used to describe the probability of the event during processing of the images captured. The software package then evaluates the processed results with a degree of accuracy and precision, including true positives/negatives and false positives/negatives, this method of identifying phenotypic variables is one of the best ways of providing accuracy and its precision is close to 100%. The clinical utility of this tool is that it can be used to perform automated screening for the existence of CTCs in patient samples and to support clinicians performing personalized medicine evaluation.
Finally, overcoming several other challenges in contemporary CTC research would result in system possessing (i) decreased assay preparation times, (ii) development of ultrasensitive biorecognition and isolation CTC platforms leading to (iii) the assembly of integrated microfluidic devices possessing (v) diagnostic modalities suitable for both single cell and clustered cell characterization. For more details on current CTCs challenges, readers are directed to the following references [1,46].

CTC technologies
Emerging platforms with nanomaterial interfaces are showing greater utility in electrochemical and optical sensing than traditional bulk materials. As most next-generation methods are derived from well-established nanotechnology approaches, modern microfluidic devices are increasingly becoming more reliable in their application as CTC isolation/enrichment and CTC recognition platforms [49,[62][63][64]. For example, previous studies have reported nanowire detection for prostate cancer (reported limit of detection (LOD): 1 CTC per 10 ml blood) [65] and the use of SERS, for head and neck cancer detection (LOD: 5 CTCs ml −1 ) [66]. In this review, we highlight CTC methodologies under the broadest of classifications. Tables 2-4 provide a comprehensive summary of the various CTC platforms along with their clinical relevance.

CTC isolation/enrichment platforms
CTC technologies in this category include physical separation and microfluidic-based platforms which have been extensively reviewed recently [57,64,[67][68][69][70][71][72]. The methodologies are based on CTC differentiation from normal haematopoietic cells based on physical characteristics including size, density, electric charges, deformability, buoyant density and the dielectrophoretic mobility of cells in varying macro/microenvironments. Microfiltration devices are the most common in this group. Recently, a thermal nanoimprint method using polyethylglycol (PEG) has been applied to develop a microfiltration CTC trapping device in which blood is filtered through pores of defined sizes, thereby enabling cells larger than the pore size to be trapped. This type of device facilitates rapid isolation of CTCs without the need for labelling. The developed method achieved an average capture efficiency of 84% for lung cancer cells spiked into blood samples [73]. Although this method is attractive, the isolated CTCs are mostly impure or of mixed origin owing to the heterogeneity of cancer cells. Therefore, most technologies in this category are encumbered by size-selection issues. Additionally, detachment of CTCs from surfaces results in damaged and clustered CTCs cells arising from haemodynamic stress. A list of devices available on the market have been described in a recent review article [74] including Ficoll OncoQuick™ [75], CellSieve™ (Rockville, MD, USA) [76], ISET ® (Paris, France) [77], Cluster Chip [78], Parsotrix™ (Angle PLC, Guildford, UK) [36], Resettable Cell Trap [74], Flexible Micro Spring Array (FMSA) [79] and ScreenCell (Sarcelles, France) systems.
Among CTC enrichment-based technologies, the Ficoll OncoQuick system is dependent on density gradient centrifugation and is based on positive selection. However, the Ficoll OncoQuick system is known to result in substantial CTC loss [7,75]. Dielectrophoretic field-flow fractionation (DEP) isolation approaches are also enrichment-based methods. They can be used to differentiate blood cells by differences in their dielectrophoretic properties. The technology can process millions of live cells in 30 min, with a high recovery rate. In practice, the platform combines the functions of DEP, as well as uses sedimentation and hydrodynamic lift forces to impose differential forces along the flow of cell types in the mixture. The platform has been used effectively in antigenindependent isolation of CTCs from blood [80][81][82][83].

CTC recognition platforms
CTC recognition platforms are based on the affinity of the specific antibodies used for quantifying the characteristics of biological processes toward antigens on targets (biomarkers). The measurable outcome may not necessarily correlate with a patient's experience but may be considered as a surrogate endpoint if there is consistency, and a clinical outcome is accurately predicted [84]. Therefore, biomarkers are considered both provisional and surrogate endpoints in clinical settings [84]. EpCAM is the most frequently used cell surface marker for the positive enrichment of CTCs, whereas cytokeratins (CKs) are targeted for post-enrichment specific to CTCs. This immunological procedure has been used in a range of techniques, including the only FDA-approved platform, CellSearch. EpCAM is a type I glycosylated membrane protein (30-40 kDa) that is overexpressed in most solid carcinomas but has low expression in healthy human epithelial tissues [85]. It has also been detected in benign colon disease [86]. However, EMT in cancer cells may lead to downregulation of both EpCAM and CKs, thus resulting in the exclusion of CTCs with a mesenchymal phenotype, which express high levels of mesenchymal markers [14,87]. As an alternative to N-cadherin, other antigen-expressing phenotype variables have been found to increase the biorecognition performance of the CTC platform when a cocktail of antibodies for various epithelial and mesenchymal markers royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 is used [14]. The negative depletion of blood cells is also a promising alternative to marker-based positive selection for CTCs that has been widely used with antibodies against CD45 to capture and deplete leucocytes [88].
The biorecognition platforms in this section consist of both functional and nanomaterial-based assays (table 3). Functional assays specifically target proteins expressed by tumour cells. Examples include collagen adhesion matrix (CAM) assays and epithelial ImmunoSpot assay (EPISPOT). Whereas the CAM assay involves cell digestion, EPISPOT assays remove leucocyte antigens via CD45 depletion; the above-mentioned two methods have high specificity and selectivity. Additionally, EPISOT assays have inherent purity issues [67,89]. Other examples of EpCAM-based recognition technologies apart from the FDA-approved CellSearch have also been reported. These EpCAM-based technologies include MagSweeper [90], which provides high purity cells that may be further used in western blot assays; GILUPI CellCollector, which is mostly used in in vivo collection of very large sample volumes [91], and CTC-iChip [92], in which target cell sizes are positively/ negatively enriched by using an antibody-dependent cell capture method. Herringbone Chip [93] and CTC chip (also known as microposts array, figure 4f ) [94] are based on the principle of microvortices, which involve the passive mixing of blood samples to significantly increase interactions between target CTCs with an anti-EpCAM-coated chip. Anti-EpCAM can be replaced with another antibody for negative enrichment. The Adna Test [95] is used mainly for molecular profiling of HER2 and EpCAM. IsoFlux, which is also a microfluidic platform, uses flow control and immunomagnetic capture to enhance CTC isolation and immunofluorescence detection for counting individual CTCs [96].
Nanomaterial-based assays are increasingly used in the diagnostics field [97][98][99][100]. These materials have a robust structure, small size (nanometre) and controllable pore sizes. They also have a high surface-to-volume ratio and are chemically tailorable to specific dimensions. Moreover, assay-based magnetic nanoparticles (MNPs) have been used directly to capture cancer cells from a working buffer solution or spiked cancer cells in phlebotomy samples. In a most recent study, nanozyme containing iron oxide loaded with gold particles was used for cancer cell capture, cancer profiling of different markers, and in post capture expansion in vitro study [101]. The nanozyme platform showed a fast response time within a linear dynamic range, 10-10 5 cells ml −1 for investigated breast carcinoma of T47D breast cancer cell line, greater than 80% capture from 10 2 to 10 5 cells ml −1 , and LOD was calculated at 0.4 U ml −1 , with limit of quantification, 4 U ml −1 in sterile PBS buffer solution. The advantage of magnetic nanomaterials is that captured CTCs can be purified after capture with an external magnet to decrease non-specific adsorption or noise before they are transferred to the detection platform. Thus, nanomaterialbased biorecognition technologies have been used to provide  MagSweeper EpCAM + device can process up to 9 ml of blood per hour. The device can also be used to enrich target cells, eliminate unbound cells and the process does not perturb gene expression of cells [90] AdnaTest EpCAM + , gene markers by RT-PCR in addition of AdnaTest being complementary to EpCAM, it also has wide clinical applications for gene detection, multiplex-PCR and fluorescence in situ hybridization [95] IsoFlux EpCAM + , MUC1, Mesothelin high recovery cell enumeration rate as compared to CellSearch: 95%: 36%, respectively; also used for tracking oncogene mutational changes [96] GILUPI CellCollector EpCAM + in vivo CTC enrichment for monitoring patient. The device is portable and safe when used as a functionalized wire [91] EPISOT assay epithelial secreting cells and CD45 + depletion device can be used to screen viable epithelial cells. The device is unbiased towards CTC/DTC phenotypes enrichment [89] CAM assay EpCAM + CAM assay favours the isolation of CTCs with collagen invasive phenotype [67] ELISPOT cells secreting specific antibody more reliable for ex vivo simulated PBMC than ELISA [177] Vita-Assay™ functional capture of CAM + successfully used for ex vivo drug-sensitivity testing of ovarian-cancer patient CTCs other applications include multiplex cytometric detection in breast, ovary, prostate, pancreas, colorectum and lung cancer [178,179] µHall detector assay simultaneous EGFR + , HER2 + and EpCAM + it can be used for clinical enumeration of rare cells, with the possibility of detecting multiple biomarkers. The sensitivity of the device as a cell counter is better than existing devices, making it a potential device for CTCs enumeration count and detection of TNBC in cancer patients [180] Ephesia CTC chip EpCAM + the technology combines immunomagnetic capture and the best aspects of microfluidic features. It is compact, fully automated, high purity capture, with sample retrieval for further genetic profiling the device has potential to non-invasively monitor CTC progression and drug-resistant mutation [181] Vortex micromixers EpCAM + ,CD45 + mutli-vortex micromixing can be used for improving binding interaction between antibodies and MNP. The magnetic sorter increases selection for targets captured. The device has the optional route for depleting WBCs and enriching heterogeneous CTCs population with minimal CTCs loss [182] FASTCell CK, CD45 − , DAPI the device can be used to locate immunofluorescently labelled rare cells on glass substrates, with scan rates ×500 faster than automated digital microscopy [183] royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 Table 4. Comprehensive summary of CTCs and technology platforms. Abbreviations: 2D and 3D, two-dimensional and three-dimensional. These notes have been compiled from the following references [38,46,[191][192][193][194][195].
CTCs challenges opportunities strengths weaknesses CTCs are rare, present in extreme low concentration in blood amidst millions of red blood cells and billions of white blood cells [184,185] [48,189] development of novel nanomaterials for performing highly multiplexed detection and molecular profiling of CTCs heterogeneity and triple negative breast cancer research will also help reduce cost medical costs integration of imprintable theranostic magnetic core including plasmonic shell star shape nanoparticlebased assays into paper microfluidic technology for multiplexed detection demonstrate some of the attainments of CTCs technology most microfluidic platforms present 2D model solids.
However, with imprintable 3D and 4D a mechanoresponsive environments, microfluidic devices and cell culture model expansion for culturing CTCs heterogeneity will lead to unprecedented breakthroughs, and better understanding of CTC characteristics preclinical studies use cells grown on 2D culture that are mostly flat, whereas a 3D culture grows upward in the z-direction, known to mimic the physiologic cell environment. This may also contribute to the reason why most drugs fail during the preclinical phase [190] other areas of CTCs technology development should focus on critical clinical relevance barcoded assays integrated with either metallic or micro/nanomaterials for multiplexed disease testing and diagnostics is also another technology that is rapidly increasing. These devices have the capabilities of broadening encoding barcode libraries information for multiplexed assays for batch processing for simultaneous detection and thereby reducing time and cost for detection of biomarker. Strategies for assembling the device enables room for: multiplexing, optical encoding, molecular tags and imaging, diversity in colour visualization for genetic encoding and graphical encoding, along with the whole range of other encoding techniques the comprehensive profiling of CTCs either directly from cultured cells or from patient samples is urgently required as this represents cancer aetiology a 4D printing has just recently been coined in printing technology [196][197][198]. It represents the ability of materials to exist in two stable forms in response to environmental changes such as solvents, heat, mechanical stress or other external stimuli that effect change. Importantly, it is expected that in the stable configuration of the material, the degree of freedom must be zero [198]. Consequently, using such materials for microfluidic systems and as biocompatible scaffold materials are breakthroughs envisaged in the future.

Microfluidic platforms
Microfluidic platforms are miniaturized devices capable of isolating rare tumour cells with tailored microfilters of varying sizes, that consider flexibility, size, control, geometry and density with precision [92][93][94]. These platforms are used for analysing blood samples within a short time, with a high recovery rate and high throughput. They can function as a microcentrifugation device and can also be integrated with other nanotechnologies. The surface chemistry of these platforms allows the design of further functionalities for the controllable release of cells. Microfluidic platforms have also been used in situ for the sequential analysis of isolated CTCs [93][94][95][96][102][103][104][105][106]. Microfluidics have been tailored as micromixers [107] to create chaotic advection at a low Reynolds number (300 µm s −1 ) or induce electrokinetic flow within a channel, the latter being used for translocating cells in low-cost microfluidic cytometry for tumour cell detection and enumeration [108]. A microfluidic isolation/ enrichment platform that features state-of-the-art technology is a multifaceted, integral platform with multi-orifice flow fractionation and a dielectrophoretic (DEP) cell separation device for continuous flow separation of cancer cells from blood samples [109,110]. Another example in this group is the microfluidic labelfree approach, which presumably circumvents biomarker heterogeneity related to CTC loss. The device also provides an unbiased CTC enrichment, despite the average size overlap between physical properties of CTCs and other blood cells that might be present. Recent studies have reported that smaller CTC sizes below the pore size of the membrane filter (8-11 µm) are cut-off, thus accounting for the majority of CTC loss during isolation [104].
Although microfluidic platforms present a variety of ways for capturing CTC at the single-cell level, the quantification and classification of CTCs remains challenging in modern biology for sorting of cells. One of the most advanced microfluidic cell sorting technologies is an automated, high-throughput enrichment CTC identification/enumeration system known as eDAR (ensemble-decision aliquot ranking) [111][112][113]. The principle of the eDAR workflow is automatic sorting by size differentiation of CTCs within an aliquot ensemble. After the CTCs are sorted into various sizes, they are transferred to a microfluidic platform for further purification. The device is all-encompassing in that the workflow involves four enterable automated loops for (i) aliquot ranking, (ii) aliquot sorting and (iii) on-chip purification. Subsequent steps within the loop may involve labelling with secondary antibodies for downstream assays before image analysis. The microfluidic device is therefore considered useful for single-cell analysis and detection. The working principle for microfluidic systems reviewed here is explained in the following sections and depicted in figure 4. Microcavity arrays for CTC entrapment which has a section of the microfluidic device integrated with a size-selective platform for rapidly isolating CTCs from other cell types (haematological cells) in whole blood, on the basis of size differences and cell deformability are illustrated in figure 4a. The recovered CTC cells on a microcavity array are amenable to downstream fluorescence profiling of immunophenotypic cells through scanning-automated fluorescence. This device can efficiently trap single cells, including stem cells and progenitor cells. The attractive feature of this device is the customizable geometry which can be optimized to provide precisely controlled separation based on cellular size differentiation from whole blood. In addition, entrapment of CTCs does not require the expression of epithelial markers to analyse the number of trapped cells. Instead, the numbers are determined primarily by scanning and counting the specified area under an automated fluorescence microscope.
The spiral biochip depicted in figure 4b is an enrichment, label-free microfluidic CTC device that allows size-based separation of viable CTCs by using hydrodynamic forces present in curvilinear microchannels [114,115]. The curvilinear channel is governed by Poiseuille flow conditions within the curved microchannel wall, where concentration rearrangements of particles occur. Particles of variable (or fractionable) sizes tend to equilibrate at the microchannel cross-section located at the different spiral bends, where they are acted on by inertial lift and Dean drag forces. The exertion of inertial lift force leads to particle migration from the channel centre while the particles within the curvilinear channel experience centrifugal acceleration and are directed radially outward, thus forming two counter-rotating vortices known as Dean vortices and also known as secondary flow [116,117]. The magnitude of this flow is quantified by a dimensionless Dean number (De): where ρ is the density of the fluid medium (kg m −3 ), U f is the average fluid velocity (m s −1 ), D h is the diameter of the channel (m), µ is the fluid viscosity (kg m −1 s −1 )), R is the radius of curvature (m) of the channel, and Re is the flow Reynolds number. Equation (2.1) is relevant for fluid flow within a curved channel when the inertial force is significant. Therefore, faster moving fluid near the walls re-circulates inward to create two counter-rotating vortices perpendicular to the primary flow direction [117]. These counter-rotating vortices, which give rise to secondary flow in microfluidic systems, have been used in fluid mixing on the basis of the concept of chaotic advection to create diffusive mixing in three-dimensional twisted channels and spiral curved channels. These systems (or vortices) result in efficient mixing and significant improvement over planar serpentine channels providing advantages for system [117][118][119].
Other label-free microfluidic CTC devices that allow sizebased separation of viable CTCs are the straight microchannels described from the Papautsky lab [116,120]. The latest device from the laboratory is the straight microfluidic chip using a multi-flow channel [121]. A straight microchannel device is based on the working principle of inertial focusing that allows migrating cells to primarily flow through the microchannels until encumbered by sidewalls within the microchannel creating a shear-induced lift force. Consequently, cells gently migrate to equilibrium centred positions relative to the sidewalls under the influence of the rotation-induced lift force. Equilibration of particles positioning within a flow is managed by (1) shear-induced lift force, (2) wall-induced lift force and (3) rotation-induced lift force [122]. Migrating larger cells are closer to the channel centreline than smaller cells in a rectangular cross-section microchannel within an observation window frame [123]. However, in the multi-flow channel device developed from the Papautsky lab, the lateral migration of cells mainly depends on the size of the cells. This process allowed CTCs to be separated from WBCs without the use of affinity-based approaches. Also, with the latest device, the Papautsky lab achieved (i) (greater than 87%) purity separation of CTCs from WBCs without labelling, (ii) 93% recovery rate at clinically relevant concentration of spiked cancer cells, (iii) the device also detected six CTCs from NSCLC patients, (iv) non-recovery of cells from five healthy control subjects. With this latest device, the authors identified a potential application in CTCs/CTM separation, which may be extended to the separation of other cancer cell types.
Another platform with a particle trapping mechanism is the vortex chip (figure 4c). This technology is based on two working principles: (i) inertial focusing and high-throughput particle alignment, and (ii) laminar microvortices and a particle trapping method. The first principle uses the alignment of randomly dispersed particle flow through a channel in a manner dependent on the Re value. If Re ≥ 1, two counteracting inertial lift forces are immediately in place: a shear-gradient lift force, F LS , which directs particles toward the channel walls, and the concomitant wall effect lift force, F LW , which repels the particles toward the channel centreline, which by default branches to two to four dynamic equilibrium positions located between the channel centreline and the wall. The second part of this technology is the expansion of figure 4c that represents the entrance into the reservoir [124]. In principle, the reservoir makes use of multiple expansion-contraction for predictable laminar vortices known as Moffatt corner eddy flow [125], in which the flow near the corner between plane boundaries and a rigid wall at rest is determined by a certain critical angle. When the critical angle is lower than the angles between the planes, a series of eddy flows arise, decreasing in size and intensity. The practical application of this principle to cells of varying sizes within the reservoir is that larger cells are pushed away from the channel centre through a separatrixa boundary that divides two modes of distinct flow behaviour into vortices-and remain stably trapped.
Another entrapping device known as a cluster chip is shown in figure 4d. The device, by design, captures cluster cells that flow through the triangular pillars, streaming cells under a dynamic force balance at the cell-cell junctions to support a stable equilibrium [78]. The design is based on the juxtaposition of rigid triangular planes in arrays, such that the triangular vertices are in proximity, with an opening of approximately 12 µm × 100 µm. Under a low shear stress of blood flowing through the channel, only CTC clusters are captured, and single cells pass through. This device allows not only the capture of CTC clusters without tumour-specific markers from unprocessed blood but also integrity of cells captured is preserved. The Cluster chip device is also one of the few CTC platforms that exploits the geometry of aggregate cells to differentiate them from single cells in blood. Additionally, the device allows the downstream applications of CTC culture expansion.
A tailored filtration device in which the sample is placed on a Teflon membrane surface, with strong ring magnets (not shown in the diagram) for magnet-assisted sealing and a waste outlet is shown in figure 4e. Parylene C is (a substitution of Teflon membrane) used for cell filtration, which has a micropore membrane that can be arrayed with different pore diameters (8-12 µm), is easily handled with tweezers. The mechanical strength of the membrane, high porosity and high reliability in liquid biopsy handling have been reported [126]. The high porosity of the membrane has been attributed to the edge-to-edge face of the micropore, increasing the high recovery rate of viable cells. Simultaneously, non-specific adhesion arising from background cells is decreased [127]. The micropore array platform design achieves a higher CTC recovery rate of 80-96%, compared with similar filtration structures [127]. In addition to the use of this platform for CTC separation, the device has found broad applications in separating large volumes of exfoliated tumour cells from other body fluids such as bronchoalveolar lavage fluid, (40-80 ml), urine (500-1000 ml) and pleural fluid (500-1000 ml), and has also been used to obtain more comprehensive information about ex vivo tumour sites; consequently, it is expected to be beneficial in precision medicine in the future [126,127].
The Lim group [128] has recently developed a new platform called Microfluidic Centrifugal Nanoparticles Separation and Extraction (µCENSE, figure 4g). The device has been used for rapid, label-free isolation of extracellular vesicles, EVs [129]. EVs play a vital role in intercellular communication among membrane cells, lipids and RNA [130] and are regarded as important biomarkers. The µCENSE technology is based on the principle of centrifugal microhydrodynamics [131] for separating particles according to their size differences. Pressurized fluidized particles driven along the serpentine separation channel are pushed away by radial particles toward the centre of rotation based on size-selection. The centrifugal force that generates spin for the rotor assembly is actuated by a stream of air rushing into the inlet, which is set to operate at a given duration. The microfluidic chip is then removed and observed under a fluorescence microscope. Downstream analysis can be monitored by extraction of microparticles under vacuum. The platform potentially demonstrates EV extraction on a table-top centrifuge within approximately 8 min, with 85% purity. The device is considered clinically relevant in point-of-care diagnostics because it does not require the use of a syringe pump and other accessories that might require a professional handler.
ClearCell FX system Biolidics Limited (formerly, Clearbridge BioMedics) has recently developed a simple microfluidic chip used for downstream analysis. While this invention is relatively new and is patent, it has wide application for diagnostic, personalized treatment and prognostic evaluation. It has been shown to be compatible with a label-free inertial microfluidics device for CTC enrichment in this study [132] for the examination of 30 patients with head and neck cancer, esophageal cancer, gastric cancer, colorectal cancer in the first phase, including advanced colorectal cancer in the second-phase examination.
Another device using a liquid biopsy method is the labon-a-disc, Exo-Hexa table-top system (figure 4h). The lack of efficient method for androgen-receptor splice variant 7(ARV-7) CTC test associated with castration-resistant prostate cancer (CRPC) remains a challenge in clinical practice. Extracellular vesicles (EVs) are secreted in most bodily fluids and hold promise as a non-invasive method. Particularly AR-V7 has been identified in advanced stage prostate cancer patients [133]. A recent discovery highlights the use of a urine-based liquid biopsy device for prostate cancer detection [134]. The working principle of the Exo-hexa disc device, which is a modification of the previous Exo-disc [135], is that it has three chambers. The first chamber is used for sample loading, the second chamber for filtration and the third chamber as a waste storage, positioned in a radial outward direction (figure 4h). The centrifugal pumping force is the mechanical force that drives the filtrate through the membrane and is tangential to the flow direction. Once the filtrate is collected, it undergoes downstream analyses using quantitative room temperature, polymerase chain reaction (RT-PCR) for mRNA extraction and ddPCR system for the simultaneous extraction of AR-V7 and AR-FL mRNA [134].
Notably, Peter Kuhn's lab developed the first AR-V7 CTC test for prostate from liquid biopsy using Epic CTC while he was at Scripps Research Institute in La Jolla, California [136] . The significant advantage of Kuhn's method is that it offers the unbiased approach of plating all 300 million cells on a blood draw on a slide, and combined with advanced information technology, grouped every individual cell on the slide to their respective coordinates (X,Y) [136]. The Epic platform also has the capability of providing comprehensive profiling of CTC, integrate with other existing technologies such as CellSearch, and downstream analyses of single cell at the genomic resolution level [136,137].

Techniques based on electrochemistry, surface plasmon resonance and surface-enhanced Raman scattering
One of the attractive means of identifying and quantifying a bioanalytical target is direct conversion of a biological event to an electronic signal. Commonly used techniques for detection are cyclic voltammetry or differential pulse voltammetry, chronoamperometry, chronopotentiometry and impedance spectroscopy. Therefore, electrochemical techniques provide a valuable resource for rapid quantification of DNA, RNA, cancer cells, exosomes and so forth. SPR or SERS are quantification techniques that are sometimes used as a complementary detection platform. Electrochemical detection techniques are extremely useful tools for analysing the redox state of chemical or biological systems, and third-generation biosensors are increasingly explored for their high market potential, superior sensitivity and nanoporous-mimicking enzyme fabrication for enhancing signal amplification [138]. SPR uses light to interrogate the interface nature of surface platforms and measure the adsorption of dynamic processes on thin films of gold, silver or metal nanoparticles. It provides the basis for many colour-based biosensor applications [139,140]. As an interface technique, SPR uses a monochromatic light on a surface to activate the charge-density of electrons (electrostatic potential) from modified metal surfaces. Examples of metal surfaces considered to be good candidates for optical excitation of surface plasmons include gold, silver, copper and aluminium. An interface with a dielectric medium is required for electrons to be available on a metal surface. A practical way of generating surface plasmons is by using attenuated total reflection, rather than using a direct reflection technique or using a Kretschmann prism configuration [141,142]. Therefore, real-time SPR applications for biospecific interaction analyses require (i) a biosensing surface immobilized on a metal surface as a coupling matrix, (ii) SPR optics for the convergence of incident light, (iii) a computer with an appropriate software package for the determination of the resonance angle, presentation and handling of data, and (iv) a simple microfluidic platform with a sample injection unit, rinsing and a regeneration unit and all units must be fully automated to achieve a real-time SPR measurement [143].
Similarly, SERS is also an interface technique used for amplifying the Raman signals from molecules by several orders of magnitude. The amplification mainly comes from the interaction of electromagnetic light with molecules adsorbed onto metal surfaces, thereby producing large amplifications generally known as plasmon resonances [144].

Nanoparticle-based electrochemical/SPR/SERS technologies for CTC application
Wang et al. [145] have recently used a combined EC/SPR technique for label-free capture and detection of CTCs. They modified a glassy-carbon electrode with a plasmonic gold nanostar and then immobilized an aptamer probe directly on the surface. Based on direct plasmon-enhanced electrochemistry, they were able to observe hybridization of plasmon concentrations at the star-shaped tips, producing an increased excitation of the nanoparticle material core and an enhancement of the electromagnetic field at the star-shaped tip. Based on this concept, the Wang group has immobilized the aptamer probe on a gold nanostar modified glassy-carbon electrode to selectively capture the CTCs in blood samples, thus markedly enhancing the detection signal readout.
A multifunctional-based plasmonic application for photothermal destruction of CTCs has also been reported by Fan et al. [100]. Construction of the device required functionalization of a MNP 'core' with a gold coating (as used for most aptameric modified surfaces) and attachment of a protein-specific aptamer for subsequent fluorescence imaging. The authors achieved isolation of specific cancer cells from the sample after attachment to the magnetic core with the aid of an external magnet. Cell death was initiated by irradiation of the magnetic beads with light of a specific wavelength whereupon the resultant localized optical heating resulted in cancer cell death (in that study: SKBR-3 cells). To confirm the experiment, dead cells were successfully stained with trypan blue-a dye that binds only to dead cells (and appears colourless under a bright-field microscope). Thus, this method provides a highly selective, simple recognition platform that not only rapidly and efficiently isolates cancer cells but also can be used for imaging and qualitative determination of cell viability. The same group previously reported similar work using functionalized magnetic coreplasmonic shell star-shaped nanoparticles for isolation of CTCs from whole blood samples [99].
Recently, a paper-based SERS device that also uses magnetic separation techniques has been used for cancer screening of colon cancer cells, normal cells and RBCs present in a mixture. Each cell type was differentiated with high royalsocietypublishing.org/journal/rsif J. R. Soc. Interface 17: 20200065 sensitivity and specificity, and the capture efficiency of cancer cells was approximately 98% [146]. The plasmonic paperbased process is illustrated in figure 5. The plasmonic paper was fabricated by simply dipping laboratory filter paper in gold nanorod solution. In a parallel experiment, an immunomagnetic conjugate (anti-EpCAM functionalized with magnetic beads) for cancer cell capture was purified with an external magnet, after incubation of a mixture of cancer cells, fibroblasts and RBCs. The enriched cells containing immunomagnetic conjugates were then transferred to the gold nanorod functionalized paper plasmonic material. Excitation of the paper surface with visible light (at 633 nm) produced SERS enhanced Raman spectra 'fingerprints' for the different cells. Another example of paper-based SERS for cancer detection involves a paper matrix impregnated with silver nanoparticles which has been used for clinical diagnosis of human papillomavirus-an infection associated with cervical cancer [147]. Therefore, paper-based SERS approaches for CTC detection are cost effective, disposable, have minimum environmental impact and are mechanically advantageous as they have a large surface area and are easily adopted for applications in microfluidic devices [146].

Conclusions and future perspectives
CTCs' extreme rarity, heterogeneous nature and varied phenotypicity remain major challenges for the development and application of new detection and therapeutic methods. Although the current state-of-the-art CellSearch technology remains the only FDA-approved test for clinical CTC enumeration, the technology has been unable to address many biological challenges associated with CTC characteristics. Issues include: (i) CellSearch cannot address CTC heterogeneity and phenotype variants requiring multiple markers for isolation and detection, (ii) the premise is built mostly on EpCAM as a universal biomarker for CTC enrichment, (iii) the method performs CTC enumeration regardless of whether CTC are damaged, single or clustered, and (iv) in the absence of an algorithm or software for differentiating phenotype variables, a CTC mixture will always be present in the count. Despite these drawbacks with CellSearch technology, integration with Diagnostic LeukApheresis has recently been reported to increase CTC yield and improve molecular tumour characterization [148]. Because most clinical CTC enumeration counts are not accurate, prognostic evaluation of patient samples remains largely conjectural. Thus, there is a critical need to reevaluate patient prognosis before and after treatment. Interestingly, several studies have examined CTCs along with other associated circulating markers, including circulating cell free DNA, exosomes and proteins within a given blood sample. Such a comprehensive platform provide means for understanding diagnosis, prognosis and therapeutic treatment in disease management. There is also a critical issue regarding isolating and enriching low CTC concentrations in enough quantities without compromising their integrity during capture with a CTC device.
Integrated microfluidics developed for non-commercial and commercial systems are promising technologies for CTCs detection that have showed higher separation efficiencies. Though a few of these promising platforms still need improvement in cell clogging issues and low throughput.
Overall, microfluidic systems integrated with advanced computer systems may advance clinical research area rapidly.
The Epic system is one of the most promising enrichment platforms. From a recent study [137], the Epic technique was used to demonstrate unbiased characterization and zero detection of 18 healthy donor samples, including establishing precision between multiple operators and using staining slide batches. The most attractive feature of the Epic platform is that it is amenable for downstream analysis including the analysis of single-cell resolution.
Nanotheranostics for cancer outpatients is also envisaged to be the precision medicine of the future because it enables a single platform to be used in diagnostic, prognostic and therapeutic treatment with accuracy. Interestingly, several paper-based SERS applications are promising candidates for rapid CTC detection, especially in remote, developing countries. A comprehensive summary highlighting CTCs challenges, opportunities, strengths, weaknesses, and CTCs technology platforms is in table 4.

Note
The concept of tumour cells translocating from a primary mass and invading locally in blood circulation was reported by Récamier [200]. Langenbeck's [201] observations provided the first experimental evidence of tumour cells in blood circulation. Rudolph Virchow [202] proposed the cellular origin of cancer, and Thiersch [203] developed the theory of growth energy imbalance in cancer. Australian physician, Thomas Ashworth [204], documented the first evidence of epithelial cells in the blood of his deceased cancer patients, and consequently discovered CTCs. Three decades later, Stephen Paget [205] proposed the seed-and-soil hypothesis of metastatic disease. In a series of autopsy studies, James Ewing [206] showed the circulatory pattern to be the vehicle that controls primary tumour spread to secondary organ site, for example, breast cancer metastasizing to bone, liver, brain and lungs, or colorectal cancer (colon origin) metastasizing in the liver. Other investigators also reported cases of metastasis not necessarily involving invasion of the vascular wall of the primary tumour and target organ. Watanabe Satoru [207] discovered CTC clusters in jugular veins of mice after the injection of bronchogenic carcinoma cells. It was not until the 1990s that the clinical relevance of CTCs, along with other free circulating biomarkers in blood samples (liquid biopsy) received recognition as alternatives to frequent bone marrow biopsies. The commentary notes can be found in the following references [8,27,37,.
Data accessibility. The authors confirm that all data supporting the findings of this study are already available within this paper.
Authors' contributions. V.A. participated in the design of the study, coordinated the study and draft the manuscript; T.H.K. coordinated the manuscript and helped draft the manuscript; C.L.B. coordinated the manuscript and helped draft the manuscript; I.E.C. coordinated the study and helped draft the manuscript. All authors gave final approval for publication.
Competing interests. All authors declare no competing interest. Funding. V.A. acknowledges research grant support from the Griffith University towards a doctoral programme.