Colorectal cancer diagnostics: biomarkers, cell-free DNA, circulating tumor cells and defining heterogeneous populations by single-cell analysis

Abstract

Reliable biomarkers are needed to guide treatment of colorectal cancer, as well as for surveillance to detect recurrence and monitor therapeutic response. In this review, the authors discuss the use of various biomarkers in addition to serum carcinoembryonic antigen, the current surveillance method for metastatic recurrence after resection. The clinical relevance of mutations including microsatellite instability, KRAS, BRAF and SMAD4 is addressed. The role of circulating tumor cells and cell-free DNA with regards to their implementation into clinical use is discussed, as well as how single-cell analysis may fit into a monitoring program. The detection and characterization of circulating tumor cells and cell-free DNA in colorectal cancer patients will not only improve the understanding of the development of metastasis, but may also supplant the use of other biomarkers.

Keywords::

• biomarkers
• cell-free DNA
• circulating tumor cells
• colorectal cancer
• colon cancer
• microsatellite instability
• plasma DNA
• single-cell profiling

Colorectal carcinoma is the third most common cancer diagnosed worldwide and the fourth most common cause of cancer death [1,2]. Oncologic outcomes have improved with advances in surgical technique and adjuvant therapy. Overall 5-year survival for colorectal cancer is excellent for disease localized to the colon (95% for stage I and 82% for stage II) (Table 1), but decreases to 61% for patients with regional spread to the lymph nodes (stage III) and only 8% for patients with distant metastases (stage IV) [3,4]. Fewer than 40% of the patients in the USA are diagnosed with stage I or II disease, for which surgery alone can be curative. Patients with stage III disease, comprising 36–54%, are candidates for adjuvant chemotherapy after surgical resection. Patients with stage IV disease comprise 20–31% and usually undergo first-line chemotherapy, potentially followed by surgical resection [5]. Metastasis occurs most commonly to the liver in a synchronous fashion in 14.5% and in a metachronous fashion in 60% [6,7]. Synchronous metastasis to the peritoneal surface occurs in 9–18% of the patients, to the lung in 10–14% and to the bone in 0.7–1.6% [8]. There are several regimens used for the treatment of stage III and IV colorectal cancer, including 5-fluorouracil, leucovorin and oxaliplatin (FOLFOX); 5-fluorouracil, leucovorin and irinotecan (FOLFIRI); and capecitabine (the oral form of 5-fluorouracil, otherwise known as Xeloda^®) plus oxaliplatin (XELOX). Advances in molecular diagnostics have allowed the development of more targeted therapeutic agents to reduce the risk of metastases and to treat metastases if they do occur.

The most widely used tumor marker in colorectal cancer is serum carcinoembryonic antigen (CEA), which is helpful in the detection of liver metastases developing after tumor resection. In addition, tumor-specific markers such as KRAS and microsatellite instability (MSI) are useful for both prognosis and directing chemotherapy.

Circulating tumor cells (CTCs) have been identified as a necessary component of the metastatic cascade. Numerous methods of CTC detection have been developed. Studies demonstrating that CTCs have prognostic and predictive value have spawned interest in potential clinical applications. Additionally, single-cell profiling of CTCs has opened a new frontier in personalized oncologic therapy. Cell-free DNA has also been studied for its prognostic and predictive value.

Tumor markers in colorectal cancer

CEA is the current serum biomarker used in clinical practice, with its main utility for indicating disease recurrence in the liver. Tumor biomarkers identified in colorectal cancer tissue have been used to guide chemotherapy regimens and include KRAS, BRAF, MSI and SMAD4 [9].

CEA

Since its identification in colon adenocarcinoma in 1965, CEA has become the standard tumor marker in the clinical care of patients with colorectal carcinoma [10]. It is correlated to disease stage and preoperative CEA level is related to prognosis [11–16]. Following resection of the primary tumor, serial CEA levels are routinely obtained every 2–3 months for at least 2 years to monitor for disease recurrence or metastasis [17–19]. There is no consensus on what change in CEA level would trigger further evaluation for recurrence [20]. Increases of CEA by at least 50% has a sensitivity of 76% and specificity of 90% for predicting progression of disease in patients with metastatic disease undergoing palliative chemotherapy, and a drop of at least 30% can exclude disease progression [21].

MSI

MSI occurs due to defects in DNA mismatch repair genes, and while it is typically associated with hereditary nonpolyposis colorectal cancer, most MSI-high tumors occur sporadically [22,23]. Despite their resistance to alkylating agents and cisplatin, MSI-high tumors have better recurrence-free and overall survival [23]. In patients with stage II disease, MSI-high status was found to confer the same advantage in long-term outcomes as that conferred by stage T3 over T4 [24].

KRAS

Mutant KRAS is associated with resistance to anti-EGFR monoclonal antibody immunotherapy with agents such as cetuximab or panitumumab. Randomized trials (CRYSTAL, OPUS and PRIME) demonstrated the efficacy of EGFR inhibitors when added to FOLFOX or 5-fluorouracil, leucovorin and irinotecan chemotherapy regimens for patients with KRAS wild-type tumors, especially with significantly improved progression-free survival, response rates and R0 resection rates compared with chemotherapy alone [25–28].

However, the MRC COIN and NORDIC-VII trials subsequently failed to show the same benefit, demonstrating increased response rate without improvement in progression-free or overall survival when cetuximab was added to first-line chemotherapy for metastatic disease [29,30]. Current clinical practice for patients with metastatic colorectal cancer screens tumor tissue for KRAS mutations so that only patients with wild-type KRAS tumors receive anti-EGFR therapy.

KRAS mutations occur more frequently in right-sided tumors, low-grade tumors, and MSI-low or microsatellite-stable (MSS) tumors. It does not appear to have prognostic value for recurrence-free and overall survival [31].

BRAF

Efforts to find other tumor mutations that will allow more specific targeting of patients who will benefit from anti-EGFR immunotherapy have shown that the presence of an activating mutation in BRAF in the KRAS wild-type population is associated with poor response [32,33]. However, the CRYSTAL and OPUS trials that supported the use of cetuximab in KRAS wild-type metastatic colorectal cancer also found that while BRAF mutation status could not be used in the same fashion as a predictive biomarker for chemotherapy, it did have an association with poor prognosis [34]. Similarly, the NORDIC-VII study that did not support the use of cetuximab in this same patient population also found that BRAF mutation was a strong negative prognostic factor [30].

Right-sided colon cancers with BRAF mutation and microsatellite stability have a particularly poor prognosis with regard to overall and disease-free survival. These tumors tend to have adverse histologic features such as lymphatic and perineural invasion, tumor budding and mucinous differentiation [35]. BRAF mutation in stage II and III disease has been shown to be prognostic for overall survival in the MSI-low or MSS population [24]. BRAF mutations occur more frequently in right-sided tumors, MSI-high tumors, high-grade tumors, patients over 60 years of age and female patients [31]. In preclinical studies, vemurafenib (PLX4032), a B-Raf kinase inhibitor approved by the US FDA for use in late-stage melanoma patients with BRAF mutations, showed poor efficacy in colorectal cancer (CRC) patients with a BRAF mutation [36]. This resistance to BRAF inhibitors can be overcome by using it in combination with phosphoinositide 3-kinase (PI3K) inhibitors, AKT inhibitors or standard regimens like capecitabine and bevacizumab or cetuximab and irinotecan; this has been demonstrated by in vitro as well as in vivo animal studies [37,38].

Loss of heterozygosity & SMAD4

The SMAD gene products are transcriptional mediators in the TGF-β signaling pathway. The genes encoding SMADs are located at chromosome 18q21. Initial studies demonstrated that loss of heterozygosity at chromosome 18q was a strong prognostic factor for worse survival rates in patients with stage II and III colorectal cancer [9,39,40]. Subsequent studies have demonstrated that loss of heterozygosity is less important in prognosis than previously thought, with no association with cancer-specific or overall survival, although it is independently associated with high tumor grade and KRAS mutation [24,41,42].

However, loss of SMAD4 expression, one of several genes encoded at 18q21, has been shown to be a significant independent prognostic factor for worse recurrence-free and overall survival, particularly in patients with stage III disease [24,43,44]. Mouse studies demonstrate that loss of SMAD4 expression changes the role of TGF-β from growth suppressor to growth promoter, thus increasing the tumorigenic and metastatic potential of colorectal cancer cells [45]. Patients with stage III disease and intact SMAD4 expression with MSI were found to have similar outcomes compared with patients with stage II disease, whereas patients with stage II disease and loss of SMAD4 expression without MSI status had outcomes similar to patients with stage III disease [24]. Retention of SMAD4 expression has also been found to be a predictive marker for a threefold increase in benefit from 5-fluorouracil-based chemotherapy [46]. Loss of SMAD activity occurs in 10% of the colorectal cancers and is associated with advanced-stage disease, the presence of lymph node metastases and shorter overall survival [47].

The role of CTCs in the metastatic cascade

CTCs are defined as tumor cells that have separated from either the primary tumor or metastases, and are circulating in the peripheral blood. CTCs have been detected in almost all cancers, including colon, breast, prostate, lung, ovary, pancreas, liver, gastric, esophageal, renal, bladder, thyroid, nasopharyngeal and melanoma; they are extremely rare in patients without malignancy, and can be detected before the development of metastases [48,49]. An early step in the metastatic process is the sporadic shedding of malignant cells from the primary tumor, after which these cells must access the circulation [50]. It has been observed that up to 1 million cells enter the circulation daily per gram of tumor tissue [51]. However, tumor cell embolism does not necessarily lead to metastasis, due to a process of elimination of most of these cells, called ‘metastatic inefficiency’ [52,53]. For example, trauma from mechanical shear forces causing lethal deformation can cause cancer cells in the peripheral blood to undergo apoptosis [54–56]. The majority of CTCs undergo anoikis, which is apoptosis induced by cell detachment; cells with anoikis resistance are more likely to survive in the circulation [57]. CTCs have been shown to possess physical properties that potentially aid their ability to intravasate into the peripheral blood, such as increased cell elasticity [58]. Even after hematogenous dissemination, tumor cells are highly inefficient in the processes of extravasation and subsequent growth. A small proportion of extravasated cells might form micrometastases, most of which disappear while only a few progress to form macroscopic tumors [53,56]. This process of extravasation and growth involves the attachment of tumor cells to the vascular endothelium followed by cancer cell aggregation at these sites of attachment [59,60].

Another contribution CTCs may make to the progression of malignancy is the development of highly metastatic phenotypes due to genetic exchange between tumor cells [61,62]. The epithelial–mesenchymal transition (EMT) may also be a key step for metastatic spread to occur by way of CTCs [63–69]. Single-cell analysis has been used to identify and characterize CTCs, and in breast and prostate cancer, CTCs have been found to express EMT-related genes, with these cells more commonly associated with metastatic disease [58]. Interactions with platelets also play a role in allowing tumor cells to make the transition to an invasive mesenchymal phenotype leading to metastasis [70].

For formation of metastasis, cells undergo a reversed process referred to as mesenchymal–epithelial transition (MET) [71]. However, it is not yet clearly established whether EMT is a necessary step leading to the development of metastases as there is evidence that cell invasion can occur in the absence of EMT, and whether MET plays an important role [72,73]. Nevertheless, recent work demonstrates that a small subset of CTCs is indeed capable of inducing metastasis. These cells express epithelial cell adhesion molecule (EpCAM), CD44, CD47 and MET in patients with luminal breast cancers [74]. Further investigation is needed to fully understand this process.

As demonstrated by genomic analysis of single cells from primary and metastatic tumors, there is genetic diversity within primary tumors, and mutated tumor cells in the primary tumor can then undergo clonal expansion in the metastatic site [75,76]. There is substantial lack of concordance between the mutation phenotypes of primary and metastatic tumors in colorectal cancer, seen in up to 23% for KRAS mutations. KRAS and BRAF mutations can be identified in CTCs as well, although the mutation phenotype of CTCs has not been shown to necessarily correlate with that of metastatic tumors [77,78]. Mutational analysis of KRAS, BRAF and PIK3CA in single CTCs demonstrate heterogeneity among CTCs from the same patient, which may explain the inconsistent response to EGFR inhibitors and emphasizing the importance of single-cell analysis for identifying different subpopulations of CTCs that may respond to different therapies [79,80]. Not all CTCs are necessarily reflective of the population of cells that has successfully caused metastasis. Thus, the characteristics of CTCs may differ from those of the cells of metastatic tumors [81].

CTC enrichment & capture technologies

The laboratory definition of a CTC is an epithelial cell found in a cancer patient’s blood which has an intact nucleus and shows expression of cytokeratin, but not CD45 (a white blood cell marker) [82]. A wide variety of technologies have been developed to accomplish the feat of isolating a single CTC from approximately 5 × 10⁶ nucleated leukocytes and 5 × 10⁹ red blood cells per ml [48].

The potential sources of blood for CTC analysis in patients with colorectal cancer are mesenteric, portal or peripheral. In a large prospective study, the detection efficiency of CTCs in the central venous blood compartment and mesenteric venous blood compartment was compared, demonstrating that CTCs were found at a markedly higher rate in the mesenteric venous blood compartment than in the central venous blood compartment [83]. However, since a peripheral blood draw is much more convenient and can be obtained at different time points of treatment and disease progression, most studies analyze CTCs using this approach.

The majority of CTC detection methods require an enrichment step, which may be in the form of red blood cell lysis and/or density-gradient centrifugation for elimination of red blood cells, followed by a separation from contaminating leukocytes prior to isolation and characterization [84,85]. CTC enrichment methods for colorectal cancer fall into two categories: detection according to expression of surface proteins which can be EpCAM-based or non-EpCAM-based, and detection according to other cell characteristics such as size and morphology, physical characteristics or functional properties (Tables 2 & 3).

Enrichment according to expression of surface proteins

EpCAM is a glycosylated membrane protein initially discovered as the dominant antigen on colon cancer cells and mediates epithelial-specific intercellular cell adhesion. High expression of EpCAM has been found in most adenocarcinomas, metastases and cancer stem cells and thus is considered an optimal tumor surface antigen for the detection of CTCs [86,87]. The CellSearch^™ System (Veridex, NJ, USA) is the most widely used EpCAM-based system for CTC detection, and the only one that is FDA-approved for use in patients with metastatic colorectal, breast and prostate cancer [88–94]. For the purposes of prognostic and predictive value, the cut-off value separating favorable and unfavorable CTC in metastatic colorectal cancer is three CTCs per 7.5 ml of blood [94].

An EpCAM-based system developed by our group is the MagSweeper, which is an immunomagnetic cell separation device that isolates live cells with extremely high purity, allowing not only enumeration, but also downstream single-cell molecular characterization and growth of CTCs [95]. Multiple other systems based on EpCAM have been developed, including the Adna Test [96], microfluidic chips such as the Herringbone Chip or the newly developed iChip, which also can be used in a negative selection mode (non-EpCAM based) [97–101], the Isoflux system (Fluxion) [102], the GILUPI Nanodetector (insertable wire) [103], as well as magnetic-activated cell separation [104]. Furthermore, there are several non-EpCAM-based methods that allow enrichment of CTCs through negative selection by exclusion of CD45-positive leukocytes (Table 2) [105–107].

Enrichment according to other cell characteristics

The following methods allow the detection of CTCs by using their different size, deformability, density and electric charges compared with normal blood cells. For isolation of CTCs by size, several different filtration methods have been developed (ISET [108,109], ScreenCell^® [110], the CREATV MicroTech filter [CellSieve^™] [111] and a Parylene filter [112]) that enable rapid isolation of CTCs on removable filters, which can immediately be used for downstream analysis. Other technologies isolate CTCs according to their cellular functionality (Vitatex) [113], by microfluidics by the use of label-free biochips [114] or Dean Flow Fractionation [115], as well as by the use of physical cell properties using dielectrophoretic field flow fractionation [116,117]. The fiber-optic array scanning technology allows detailed cytomorphologic analysis of cells (Table 3) [118].

CTC detection methods

Even after enrichment of CTCs, a considerable number of contaminating leukocytes remain in the CTC specimen, so a second step is needed for identification of CTCs. Detection and characterization of single CTCs can be carried out by performing immunocytochemistry (ICC), RT-PCR or the epithelial immunospot (EPISPOT) assay.

ICC

ICC analysis is used to identify CTCs, since there often is remaining leukocyte contamination after enrichment. Most methods use antibodies against cytokeratin, CD45 (leukocyte antigen used as a negative marker), and 4′,6-diamidino-2-phenylindole (DAPI nuclear stain) for fluorescent staining of CTCs.

RT-PCR

Using RT-PCR, the gene expression analysis of at least one of the epithelial markers KRT-7, -8, -18 and/or -19 as well as reference genes (GAPDH, ACTB) confirms the existence of a CTC. Cells expressing CD45 have to be excluded [119,120]. RT-PCR has also been used to detect CEA and CK20 from peripheral blood before and after curative resection, and from mesenteric tumor drainage blood. Positive rates of marker genes in tumor drainage blood are 3.3-times higher than that of peripheral blood [121,122].

EPISPOT assay

The EPISPOT assay is derived from the ELISA and can be used after any kind of enrichment method to detect only live CTCs by detecting proteins released from epithelial cancer cells. Enriched and captured cells are cultured for 48 h on an antibody-coated membrane. Those antibodies capture released proteins, which in a second step can be detected by adding secondary antibodies labeled with fluorochromes. Apoptotic tumor cells do not secrete measurable amounts of proteins and therefore are not detected [123].

CTCs are highly heterogeneous in their surface antigens and genetic characteristics. The only device that has received FDA approval is the CellSearch System. However, because this method is EpCAM-based, it has a large drawback in its inability to capture tumor cells undergoing EMT, as the expression of EpCAM and cytokeratins, used to identify CTCs, is downregulated, while markers of EMT (e.g., vimentin and N-cadherin) are expressed. Therefore, negative selection methods by eliminating leukocytes by CD45-depletion or label-free devices might seem more favorable for maximizing CTC enrichment. In addition, the development of novel markers of CTCs such as Plastin3 that is not repressed during EMT has successfully been shown to increase the yield of CTCs. CTCs identified by this method were also shown to be prognostically significant [124].

The large number of methods for enrichment and detection of CTCs and their use in various studies with different tumor entities renders fair comparison of results, specificity, sensitivity and reproducibility impossible. Assays successfully tested on cell-line tumor cells cannot be expected to show equal results to patient blood samples due to the heterogeneity of tumor cells even within a single patient.

The use of anti-EpCAM antibodies for the isolation of CTCs in colorectal cancer is applicable as 98% of the colorectal cancers are adenocarcinomas [125]. Furthermore, in patients suffering from hepatocellular carcinoma, EpCAM-positive CTCs have a high tumorigenic potential compared with EpCAM-negative CTCs [126]. Nevertheless, a label-free device may be more effective in the isolation of CTCs undergoing phenotypic changes, and the analysis of these cells is likely to lead to a better understanding of tumor biology and the metastatic process.

Another limitation during CTC detection is the small volume of blood screened for rare CTCs in most assays. To overcome this problem, the newly developed GILUPI Nanodetector can be functionalized with antibodies, such as EpCAM, for a highly selective enrichment of CTCs in a blood volume as high as 1.5 l. Finally, CTC detection may also be affected by trauma of the tumor during resection that induces cell shedding; thus, surgical manipulation may be a confounding factor affecting the number of CTCs detected perioperatively [127–129]. This is further illustrated by the differences in CTC detection among different blood compartments within the same patient [83]. Further studies on larger patient cohorts are necessary to be able to determine which device can most effectively and reliably detect CTCs.

Defining heterogeneous populations by single-cell analysis

Single-cell profiling techniques have been developed to analyze the genome of CTCs, not only at the DNA level, but also at the RNA level, corresponding to gene expression. Comparative genome hybridization, for example, analyzes DNA copy number variation and can be performed in single cells using PCR-based whole genome amplification [130–132]. Next-generation sequencing, fluorescence-activating cell sorting and single-nucleus sequencing are other techniques that have been developed to not only enable the identification of tumor cells, but also the characterization of the genome of those individual cells [133,134].

The minute quantities of RNA within a single cell poses a great technical challenge for the sensitivity of detection. For transcriptional analysis, after converting RNA to cDNA, amplification is necessary, after which gene expression analysis using a microarray or next-generation sequencing can be performed [82,135].

In a pioneering study, Dalerba et al. demonstrated the importance of single-cell expression analysis for improving the understanding of tumor heterogeneity and clonality [136]. Cells from normal and neoplastic human colon epithelia were isolated and fluorescence-activating cell sorting applied using established markers of differentiation to distinguish mature cells on the top of colon epithelial crypts from immature stem cells at the bottom of the crypt. In a second step, single-cell analysis by gene-expression profiling was performed. By principal component analysis, the gene expression data were then divided into different groups, defining the gene expression profiles of the differentiated enterocytes at the top of the crypts and of the immature, stem and goblet cells at the bottom of the crypts. Furthermore, a colon cancer mouse xenograft model was developed by injection of a single colorectal cancer cell into the flanks of the mice. Only cells highly expressing EpCAM and CD44, which are known to be cancer stem cell markers, resulted in tumors. By performing gene-expression profiling of these tumors and comparison with the primary tumor from which the implanted cells originated, a very similar heterogeneous cell population was found. In addition, the investigation and comparison of colorectal adenoma gene expression with healthy colon tissue demonstrated that the tumor was comprised of the exact cell populations, which shows that tumor heterogeneity is due to multilineage differentiation.

In another recent study, CTCs from 15 patients with advanced-stage colorectal cancer were isolated using the CellSearch device. The genomic profiles of the primary tumor were compared with the profiles of liver metastasis and CTCs in each patient by performing array comparative genome hybridization. This revealed that CTCs had some similarities in copy number changes to the primary tumor, as well as other similarities to the copy number changes in the liver metastasis, especially in known colorectal driver genes (e.g., KRAS, APC, PIK3CA), but that they also had additional different copy number changes. By deep sequencing of the primary and corresponding liver metastasis, it was discovered that the mutations that were found uniquely in CTCs were detectable in the primary and metastatic tissue at subclonal levels [137]. Analysis of the marker profiles of CTCs and primary tumors in nonmetastatic breast cancer patients using immunohistochemistry and FISH demonstrated no correlation between the two profiles [80]. These studies emphasize the importance of isolation and characterization of CTCs, as these cells can carry mutations that have developed in the clinical course of cancer patients and can serve as liquid biopsies to monitor therapeutic response and also help make decisions concerning therapeutic regimens for an individual patient. CTCs, however, do not necessarily carry the same mutations as the primary tumor, and this discordance may explain the current variability of treatment response to chemotherapy. The differences in profile between CTCs and primary tumors in breast cancer has been demonstrated and a similar difference in colorectal cancer is likely [80].

In the near future, it will be possible to use droplet-based microfluidics for isolation and high-throughput analysis of approximately 1 million single cells in parallel, as well as enable cell growth, which could significantly facilitate single-cell profiling. The droplets are only picoliters or nanoliters in size and are compartmentalized in the microfluidic system, leading to a faster accumulation and thus quicker detection of secreted molecules [138].

Clinical significance of CTCs

CTCs have been shown in multiple studies to correlate not only with worse clinicopathologic features of disease, but also to have prognostic and predictive value. Prognosis refers to the likely course and outcome of the disease regardless of therapy, and is measured by cancer-specific survival, disease-free survival and overall survival. The predictive value of a biomarker refers to the effect of therapy on a patient’s outcome [139].

As a marker of advanced disease

The correlation of CTC levels with clinicopathologic features that indicate more advanced disease has been observed in both metastatic and nonmetastatic disease, at various sampling points before and after surgical or systemic treatment, and in both the peripheral and mesenteric compartment. CTC detection as determined by CEA/CK20 expression in both preoperative peripheral blood and preresection tumor drainage blood is significantly associated with depth of tumor invasion, venous invasion, lymph node metastasis, liver metastasis and stage [121]. CTC levels obtained from the peripheral blood after resection of the primary tumor have been shown to have a significant correlation with regional lymph node involvement and stage of disease [140]. Similarly, CTCs obtained from the venous drainage blood after curative CRC resection also correlate with lymph node positivity [141]. For patients with metastatic disease not undergoing resection, unfavorable baseline CTC levels (>3 CTCs per 7.5 ml blood using the CellSearch System) correlate with more advanced disease. Among patients with metastatic colorectal carcinoma, those with liver metastases and poorer performance status had higher baseline CTC levels [90,94]. Another marker of CTCs, Plastin3-, has recently been shown to have significant clinical relevance. Plastin3-positivity in the peripheral blood was found to be associated with clinicopathologic risk factors of greater depth of invasion (>=T3), lymph node metastasis, liver metastasis, peritoneal dissemination, recurrence rate and Dukes stage. Plastin3 expression was also detected in all patients with recurrent disease and at a higher level compared with prerecurrence levels and to patients without recurrence [124].

As a prognostic factor

Detection of CTCs portends poor prognosis in patients with metastatic colorectal cancer. A meta-analysis of 16 studies to include 1491 patients with metastatic colorectal cancer demonstrated that patients in whom circulating tumor cells are detected have a 2.5-fold increased mortality risk and a twofold increased risk of disease progression or recurrence [142,143]. A higher percentage of patients with metastatic colorectal cancer who had disease progression or death had unfavorable CTCs at 3–5 weeks after treatment than patients whose disease did not progress. Baseline CTC is an independent prognostic factor in metastatic colorectal cancer. Patients with unfavorable levels of CTCs at baseline had significantly shorter median disease-free and overall survival than patients with fewer CTCs [90,144,145]. The effect of favorable baseline CTCs was seen in significantly longer progression-free survival time for patients receiving irinotecan, as well as in a near doubling of overall survival time for all patients regardless of whether they received oxaliplatin, irinotecan or bevacizumab. This overall effect was more pronounced in patients over 65 years of age with poor ECOG performance status [144].

For patients with nonmetastatic disease (Dukes’ stage B or C), overall and disease-free survival of patients in whom CTCs were detected were significantly worse than those of patients in whom CTCs were not detected. This difference was not seen in patients with Dukes’ stage A disease [146]. In combination with lymph node staging, CTC detection 24 h after curative resection has even greater sensitivity for predicting recurrence [122].

Patients with metastatic disease and high baseline CTC levels who demonstrate a decrease to favorable CTC levels after 3–5 weeks of chemotherapy experience the same disease-free survival as those who started with favorable baseline CTC levels, although this effect was not seen in overall survival. Patients with persistently unfavorable CTC levels prior to and after treatment have significantly worse disease-free survival and overall survival than those in whom the CTCs converted to favorable levels after treatment [90,147].

Plastin3-positivity was also associated with worse prognosis. Stage for stage, patients in whom Plastin3-positive CTCs were detected had significantly shorter overall survival than Plastin3-negative patients; patients without synchronous distant metastasis had significant shorter disease-free survival. For Dukes’ stage B and C disease, Plastin3-expression was independently associated with decreased overall survival and disease-free survival, whereas clinicopathological factors such as tumor size, lymphatic invasion, venous invasion, histologic type and depth of invasion were not [124].

In addition to CTC detection in the peripheral blood, CTC detection in tumor drainage venous blood before and after resection is also associated with prognosis. CTC detection in tumor drainage blood obtained just prior to resection has been demonstrated to correlate with overall and disease-free survival [121]. CTC levels obtained from tumor drainage venous blood after resection in Dukes’ stage B and C colorectal cancer patients was also shown to be an independent prognostic factor in both disease-free and overall survival [119]. Intraoperative CTC levels appear to be affected by surgical manipulation of tumors, whereas postoperative CTC levels are more correlated to disease-free survival [143]. Another meta-analysis examining the presence of CTCs in the venous drainage of curative colorectal cancer resections demonstrated an increased hepatic metastases rate in patients with CTCs compared with those without. Similarly, disease-free survival was significantly worse in patients with CTCs in the venous drainage blood of the tumor than those without [141].

As a predictive factor (predicting response to specific therapies)

In a trial comparing maintenance therapy for metastatic disease with either single-agent bevacizumab or XELOX plus bevacizumab after induction chemotherapy with XELOX plus bevacizumab, baseline CTC counts did not predict the response to chemotherapy. However, patients with a low CTC count after 3 months of treatment had significantly higher response rates to chemotherapy than those with a high post-therapy CTC count, resulting in a longer progression-free survival and overall survival in patients receiving chemotherapy plus bevacizumab. Patients with low baseline and post-treatment CTC counts had longer progression-free and overall survival than patients with an initially high baseline CTC count which decreased after chemotherapy. As expected, patients with persistently high baseline and post-therapy CTC counts fared the worst [145].

CTC assessments have also been shown to correlate with tumor responsiveness as determined by radiographic imaging in patients undergoing chemotherapy for metastatic colorectal cancer [120]. Combining CTC counts with tumor response as determined by CT imaging may enhance the potential of CTC as a marker of response to treatment [147].

Cell-free DNA

Cell-free DNA (cfDNA) – also known as circulating tumor DNA (ctDNA), or plasma DNA – can be detected in the blood of colorectal cancer patients. There are different hypothesis concerning the origin of cfDNA. It may emerge from tumor necrosis, apoptosis or even active secretion, but the exact mechanism is unknown [148]. It has been shown that colorectal cancer patients carrying a KRAS mutation are resistant to EGFR inhibitors, for example, cetuximab and panitumumab. By analyzing the plasma DNA concentration of 28 CRC patients receiving monotherapy with panitumumab because of an initially detectable wild-type KRAS tumor, a subset of patients showed detectable KRAS mutations, which can explain the resistance acquired after a certain period of time, in most cases after 5–6 months of therapy [149]. These mutations are detectable months before imaging reveals disease progression, implying that the analysis of KRAS mutation status in plasma can be used to evaluate therapeutic response with a simple blood draw and modulate therapy accordingly [150].

The prognostic value of cfDNA detection in plasma of CRC patients has also been demonstrated, as the 2-year survival rate was 100% in patients without evidence of cfDNA with KRAS2 mutations or p16 gene promoter hypermethylation, which can be found in 40% or 20–50% of CRC patients, respectively. The patients with detectable cfDNA had an overall 2-year survival rate of only 48% and disease recurrence [151].

In a study analyzing the plasma DNA concentrations of 32 patients with advanced-stage CRC, some showed biphasic size distributions of plasma DNA fragments, which was linked to higher numbers of CTCs and higher levels of plasma DNA mutations. Interestingly, not all patients had detectable levels of cfDNA in their circulation despite their advanced tumor stage, which could be due to the short mean half-life of cfDNA of 16 min; thus, patients might have had higher levels of cfDNA prior to the blood draw [152].

To elucidate the kinetics of cfDNA release into the circulation, another group implanted colorectal cancer cells subcutaneously into rats and performed measurements of cfDNA concentrations at different time points. They found that cfDNA was detectable at an early stage, 1 week after tumor inoculation, but that the size of the resulting tumors did not significantly correlate with the concentration of detectable cfDNA. Furthermore, cfDNA could only be detected in 50% of animals that developed metastases and the concentrations did not show significant differences when compared with the ones of animals without metastases [153].

In several studies, the evaluation of methylated SEPT9 in plasma has been shown to be useful for the detection of asymptomatic patients with average risk of CRC as well as patients of all stages of CRC. This would be an ideal method of cancer detection particularly for patients who are resistant or unable to undergo colonoscopy for screening [154–156].

Despite all of these findings, the utilization of cfDNA for therapeutic monitoring in the colorectal cancer field is not yet ready for widespread clinical use, as there are large numbers of genes that can be mutated and different studies have used different laboratory techniques for their detection. There is also no correlation among the presence and concentration of cfDNA and tumor stage, and further studies are needed to elucidate the role of cfDNA and to define specific cfDNA markers that can be detected in the majority of CRC patients. Nevertheless, cfDNA has been shown to be a powerful tool for monitoring patients who undergo therapy with EGFR inhibitors and early identification of developed drug resistance in these patients. In particular, because of the multiplicity of CTC technologies that have yet to be clinically validated in prospective large-scale trials, it may be more straightforward to monitor cfDNA for a panel of mutations that may indicate therapeutic failure. Once the presence of mutations are identified that suggest it is time to switch therapies, single-cell CTC analysis may be performed to identify whether both wild-type and mutant subpopulations of cancer cells coexist that may require specific drugs targeted to the different subpopulations.

Expert commentary

Inclusion of molecular marker assessment such as MSI status, KRAS, SMAD4, as well as CTC and cfDNA detection, may refine our ability to offer prognostic information, which is currently based mainly on TNM staging. The decision to use EGFR inhibitors such as cetuximab should be informed by KRAS mutation status. Testing for MSI is currently performed on tumors based on the revised Bethesda guidelines which screen for hereditary nonpolyposis colorectal cancer [157]. However, given that MSI can occur in the setting of sporadic colorectal cancer and has prognostic and predictive potential, routine MSI testing on all colorectal cancers does have clinical use. In particular, patients with T3N1 (stage III) disease with MSI-high tumors without loss of SMAD4 expression have similar survival to those with T3N0 (stage IIA) disease. If this survival benefit is seen in patients who do not undergo chemotherapy, then it is possible that 11% of the T3N1 population may not need to undergo adjuvant chemotherapy [24]. However, genetic heterogeneity of the cells within a single tumor may explain a suboptimal response to targeted chemotherapy [76].

CEA, despite being the current biomarker used for surveillance in colorectal cancer, has no correlation with CTCs [92]. The prognostic information that CTCs offer can be added to CEA at baseline and during treatment at 2–3-month intervals, similar to CEA testing. However, we anticipate that they may eventually be used to guide selection of chemotherapeutic agents, either by characterizing CTC populations that contain different markers, such as specific mutations that predict sensitivity or insensitivity to specific drug regimens, or by performing drug testing on CTCs grown in culture or captured fresh [158,159].

For predictive purposes, CTC levels obtained at baseline and at various time points during chemotherapy treatment can correlate with disease progression or response to chemotherapy. Conversion to favorable levels of CTCs during treatment results in significantly longer progression-free and overall survival compared with persistent unfavorable levels of CTCs. Peripheral blood seems to be the best source for CTCs in colorectal cancer, contrary to what we have seen in primary breast cancer where bone marrow seems to be a better source of disseminating tumor cells.

The existence of CTCs in the peripheral blood reveals the existence of a vascularized primary tumor and perhaps metastases, but there are still many aspects about CTCs that need to be investigated to elucidate the metastatic process, and to identify the small population of cells that are highly aggressive and capable of forming metastases or cancer recurrence [160]. Only by administration of therapies that eliminate this subpopulation of cells can a cancer can be cured [161,162]. As a biopsy or resection or administration of therapy may influence CTCs levels, multiple measurements of CTCs are necessary in order to adjust the therapeutic regimen accordingly. Thus, it is still unclear whether a single CTC within a CTC microembolus is responsible for forming metastasis, or whether a CTC invades into the target organ and then releases substances that induce mutations in surrounding cells [160].

Finally, cfDNA seems to be an ideal tool for screening and monitoring therapeutic response, especially in colorectal cancer patients receiving anti-EGFR therapy in form of cetuximab or panitumumab, because KRAS mutation, as an evidence of developed resistance, can be detected in the patient’s serum. Once response failure is identified, then single-cell analysis of CTC populations may take precedence.

Five-year view

CTCs clearly have clinical relevance and prognostic significance, as demonstrated by multiple studies. An important next step is standardization of the various approaches of CTC detection, with regards not only to the assays and markers used to detect them but also to the optimal time point in relation to surgical resection. Prospective multi-institutional trials are needed to validate the use of CTCs as prognostic biomarkers and biomarkers of recurrence, much the way that CEA is currently used.

Further exploration of the predictive potential of CTCs is warranted. If high CTC levels in patients with stage I or II disease is a poor prognostic factor for disease recurrence and adjuvant chemotherapy is shown to prolong disease-free survival in these patients, then CTCs can be used as a predictive biomarker to guide therapy. Conversely, selected patients with stage III disease with very low CTC levels and thus a low risk of disease recurrence may not warrant adjuvant chemotherapy.

Obtaining CTC levels at various time points during chemotherapy treatment is a way of tracking response to the treatment before one would typically perform follow-up imaging. Thus, patients who fail to demonstrate treatment response may be spared unnecessary toxicity to ineffective treatment and greater success by changing to a different regimen. CTC levels may also identify patients that are appropriate for a treatment break [90].

CTC detection as well as genomic analysis of the individual cells can inform adjuvant therapy decisions. Single-cell analysis of CTCs has been performed in other cancers and if applied to colorectal cancer, may be rather useful not only for predicting response to chemotherapy, as well as identifying new genes against which therapy can be developed [131,132,163]. Analysis of mutations in CTCs, which may differ from mutations in the primary tumor, has potential to direct the chemotherapeutic agents used for adjuvant therapy [77]. Whereas our current practice is to use the resected primary tumor’s mutation profile to direct adjuvant therapy, the mutation profile of the CTCs may be more appropriate for directing therapy. In addition, single-cell profiling of CTCs has the potential to play a significant role in the development of novel targeted therapies for colorectal and other cancers [82].

As detection of CTCs in peripheral blood has been shown to correlate with disease recurrence, prospective studies to measure CTC levels along with CEA levels are required to determine if CTCs can be used clinically to screen for metastases.

Circulating angiogenic factors as a potential prognostic factor is another important area of study. Low levels of PIGF, which belongs to the VEGF family of growth factors and may have a role in pathologic neovascularization, are associated with poor recurrence-free survival [164].

Since CTCs are present in patients with localized cancers, CTC detection in conjunction with single-cell sequencing also has the potential to be a diagnostic or screening test for cancer. Thus, for colorectal cancer, a blood draw could even replace the invasive colonoscopy as the preferred mode for screening and surveillance [133].

Finally, cfDNA may eventually supplant CTCs as a monitoring tool [165]. For example, it has been shown that KRAS analysis can be performed in plasma and is a feasible alternative to tissue analysis, thus offering another way to predict response to chemotherapy agents [166]. Another application of cfDNA is monitoring for colorectal cancer recurrence after curative resection [167]. However, unlike CTCs, cfDNA cannot be analyzed for cancer tumor cell heterogeneity nor used for drug testing; therefore, it is likely that both tools will be used to advance personalized therapy of CRC.

Table 1. TNM staging of colon and rectal cancer, American Joint Committee on Cancer Manual, 7th edition.

Primary tumor (T)	Regional lymph nodes (N)			Distant metastasis (M)
TX: Primary tumor cannot be assessed	NX: Regional lymph nodes cannot be assessed			M0: No distant metastasis
	N0: No regional lymph node metastasis
T0: No evidence of primary tumor	N1: Metastasis in 1-3 regional lymph nodes			M1: Distant metastasis
Tis: Carcinoma in situ: intraepithelial or invasion of lamina propria	N1a: Metastasis in one regional lymph node
T1: Tumor invades submucosa	N1b: Metastasis in 2-3 regional lymph nodes
T2: Tumor invades muscularis propria	N1c: Tumor deposit(s) in the subserosa, mesentery, or nonperitonealized pericolic or perirectal tissues without regional nodal metastasis			M1a: Metastasis confined to one organ or site
T3: Tumor invades through the muscularis propria into pericolorectal tissues	N2: Metastasis in 4 or more regional lymph nodes
T4a: Tumor penetrates to the surface of the visceral peritoneum	N2a: Metastasis in 4–6 regional lymph nodes			M1b: Metastases in more than one organ/site or the peritoneum
T4b: Tumor directly invades or is adherent to other organs or structures	N2b: Metastasis in 7 or more regional lymph nodes
Stage	T	N	M	Dukes
0	Tis	N0	M0
I	T1–T2	N0	M0	A
IIA	T3	N0	M0	B
IIB	T4a	N0	M0	B
IIC	T4b	N0	M0	B
IIIA	T1–T2, T1	N1, N2a	M0, M0	C, C
IIIB	T3–T4a, T2–T3, T1–T2b	N1, N2a, N2b	M0, M0, M0	C, C, C
IIIC	T4a, T3–T4a, T4b	N2a, N2b, N1–N2	M0, M0, M0	C, C, C
IVA	Any T	Any N	M1a
IVB	Any T	Any N	M1b

Adapted from [168].

Table 2. Enrichment of circulating tumor cells by positive (using anti-EpCAM antibodies) and negative (non-EpCAM-based) selection.

According to surface proteins	Methods	Mechanism
Positive selection (EpCAM)	CellSearch System	Consists of CellTracks Autoprep system, Epithelial Cell Kit and CellSpotter Analyzer. CellTracks Autoprep: automated sample preparation system. Epithelial Cell Kit: for enrichment of cells expressing EpCAM immunomagnetically and for fluorescent labeling of cell nuclei with 4′,6-diamidino-2-phenylindole (DAPI), of leukocytes with a monoclonal antibody against CD45, which is only expressed on these white blood cells (CD45-allophycocyan) and of epithelial cells (cytokeratins 8-, 18-, 19-phycoerythrin). CellSpotter analyzer: semiautomated fluorescence-based microscope for identification and enumeration of CTCs. Several groups showed prognostic effect of CTC numbers in cancer patients using this device: patients with metastatic breast, prostate and colorectal carcinoma have been shown to have a shorter progression-free survival if a high number of CTCs are detected; in colorectal cancer more than three cells per 7.5 ml of blood (breast and prostate cancer ≥5 cells/7.5 ml of blood [88–94])
	MagSweeper	Consists of magnetic rods covered by plastic sheaths that are swept through the samples at a particular velocity and pattern, enabling the capture of EpCAM-positive cells. Preparation: red blood cell lysis; incubation with anti-EpCAM-coated beads; incubation with cell surface stainings (Alexa 488 antihuman CD45 and phycoerythrin antihuman EpCAM); MagSweeper for isolation of CTCs, three steps: Step 1: capture; Step 2: wash – rods covered with plastic sheaths are moved to wash station filled with buffer solution to wash off contaminating and unlabeled cells; Step 3: release – magnetic rods move out of plastic cover while at the same time a magnetic field is provided, so that all captured cells are released into a buffer solution containing membrane- impermeable DAPI; repetition of steps to remove further contaminating cells using Olympus inverted microscope equipped for epifluorescence: identification of CTCs as EpCAM positive, DAPI and CD45 negative cells [95] has been used to isolate live CTCs from breast cancer patients for characterization by gene expression analysis; single-cell analysis of CTCs has demonstrated significantly different gene expression patterns than usual breast cancer cell lines, as well as heterogeneity among CTCs captured from a single blood draw and among different patients [75]
	Adna test	Uses antibodies against EpCAM and MUC-1 that are conjugated to magnetic beads. MUC-1 is a cell surface glycoprotein, often associated with epithelial cancers. Prelabelling of blood samples by incubation with a commercialized bead and conjugated antibody mixture; labeled cells are then extracted by a magnetic particle concentrator; in a second step, mRNA isolated from lysed, enriched cells can be used for downstream analysis or isolated tumor cells can be stained immunohistochemically; this method for CTC isolation in combination with RT-PCR for detection of CTCs in peripheral blood of colorectal carcinoma patients yielded a sensitive approach for CTC detection; in comparison to the CEA levels from the same patients, the detection of CTCs was possible earlier than elevated CEA serum protein levels [96]
	Microchips (CTC-Chip, Herringbone-Chip, Ephesia-Chip)	CTC-chip: contains an array of 78,000 microposts within an area of 970 mm^2; allows binding of EpCAM-positive cells to microposts without need for prelabeling of blood samples; due to minimal shear stress and lack of preprocessing of the samples, a high number of viable tumor cells (ranging from 42 to 375 per ml for colorectal cancer) are captured by the labeled microposts; visualization of CTCs by molecular analysis or by immunohistochemistry: cells positive for cytokeratin-staining and adherent to the EpCAM-positive microposts are CTCs, while CD45-positive cells are defined as contaminating hematologic cells [97]. Herringbone-chip: enhanced CTC-Chip, significantly increases number of captured cells by passive mixing of blood cells through generation of microvortices; these increase the interaction of cells with the antibody-coated surface of the chip; using this chip, clusters of CTCs were observed in patients with metastatic cancers; significance of clusters is not yet elucidated, but they may have developed from intravascular proliferation of CTCs or by embolizing from the main tumor mass into the circulation [98]. Ephesia chip: combines microfluidic and magnetic cell sorting; consists of an array of magnetic traps onto which columns of biofunctionalized beads are assembled in a microfluidic channel; allows isolation of CTCs in small sample volumes; in situ cultivation of captured cells possible when cell lines are used [99]. Chip combining microfluidics and electrophoresis: enables CTC selection in high-throughput microsampling unit followed by capture and enrichment by electrokinetic manipulation; in high-throughput microsampling unit is coated with anti-EpCAM antibodies, so CTCs are immunospecifically selected out of the blood and captured in this microfluidic channel; after enzymatical release of CTCs from EpCAM-labeled surface, they are transported through a pair of electrodes to allow conductivity-based enumeration; then cells are transported and subjected to electrokinetic enrichment as CTCs are negatively charged [100]
	GILUPI nanodetector	Recently developed method for in vivo CTC detection. Consists of medical Seldinger wire functionalized with monoclonal antibody directed against EpCAM; wire is inserted through a medical venous cannula into the cubital vein (2 cm) and left for CTC capture for 30 min; total volume of blood passing the wire is approximately 1.5–3 l; after removal of the wire, CTCs can be identified by ICC method has been successfully tested on patients with advanced-stage lung and breast cancer without adverse effects [103]
	MACS^® Technology	MACS separation column exposes immunomagnetically labeled blood samples to a strong magnet; labeled cells stay on the column while other cells flow through; depending on the magnetic nanoparticles used, both positive and negative selection can be performed. Prospective study comparing the MACS system to three other cytometric enrichment methods showed that the MACS system is highly reproducible and accurate method for CTC detection in patients with metastatic colorectal cancer and also demonstrated prognostic significance of CTC detection, as patients with metastatic colorectal cancer and more than one detectable CTC had a significantly shorter progression-free survival [104]
	IsoFlux System by Fluxion	Microfluidic technology for isolation of rare CTCs after prelabelling of blood sample with immunomagnetic capture beads (EpCAM labeled) sample is loaded into microfluidic cartridge on the device; sample passes through a microfluidic channel; cell separation: in the middle of this channel, passing cells are exposed to magnetic field in a cell isolation zone. Labeled target cells flow towards magnetic field, other cells flow through the device into waste container; target cells attach to removable disk at the roof of the isolation zone, so targeted cells can easily be aspirated and used for downstream analysis [102]
Negative selection (CD45)	RosetteSep^™ – Applied Imaging Rare Event (RARE)	Non-EpCAM-based method. CD45-positive cells are crosslinked to red blood cells by bispecific tetrameric complexes of antibodies; this leads to the formation of aggregates called ‘rosettes’ with an increasing density gradient. After density gradient centrifugation, CD45-positive cells can be detected in the lower fraction of the tube and CD45-negative cells accumulate in a layer between a separation medium and the plasma [105]
	Quadrupole magnetic cell sorter	‘Flow-through’ separation device after red blood cell lysis, leukocytes are immunomagnetically labeled with an antibody against CD45; depletion of labeled cells as the sample flows through the device in a channel surrounded with four magnets [106]

CTC: Circulating tumor cell; MACS: Magnetic-activated cell separation.

Table 3. Enrichment of circulating tumor cells by utilization of other cell characteristics.

According to other cell characteristics	Methods	Mechanism
Size	ISET	For enumeration, molecular and immunomorphological characterization of CTCs. Small sample volumes up to >1 ml can be used; filtration membrane is comprised of ten wells; allows processing of 10 ml blood at a time; filter membranes with pores 8 µm in diameter after red blood cell lysis, filtration of CTCs based on larger size than other blood cells [99]. Superior detection of CTCs in patients with metastatic prostate and lung cancer using the ISET device compared with CellSearch System in prospective study; may be due to additional capture of tumor cells in epithelial to mesenchymal transition, which leads to downregulation of EpCAM expression and upregulation of other mesenchymal markers; in addition, microemboli can frequently be detected using the ISET (but not with CellSearch method); for metastatic breast cancer, the CellSearch System had superior CTC detection than the ISET [108,109]
	ScreenCell^®	Allows isolation of fixed or live cells and tumor cell clusters in only 3 min; consists of a polycarbonate filter with calibrated circular pores (7.5 ± 0.36 µm for fixed cells or 6.5 ± 0.33 µm for live cells [110])
	CREATV MicroTech filter	Uses a CellSieve^™, which is a biocompatible polymer; achieves rapid and efficient isolation of CTCs; pore diameters are 8 µm [111]
	Parylene filter	Contains 8-µm pores, has been compared with the CellSearch System and found to have higher recovery rates of CTCs from patients with metastatic breast, prostate, colorectal and bladder cancer [112]
Cell functionality	Vitatex (Vita-Cap^™, Vita-Assay^™)	Enrichment of invasive CTCs by using a functional collagen adhesion matrix (CAM): patient samples are added to plates coated with CAM; only CTCs adhere. Captured CTCs are labeled by injection of fluorescently labeled CAM (e.g., FITC); can be used for downstream analysis; those cells are capable of propagation in culture. Using this method, CTCs were detected in 28 of 54 patients with stage I–III breast cancer with an average mean CTC count of 61/ml; correlation of number of detected cells to the tumor stage, lymph node status and survival in patients with early stage breast cancer [113]
Physical and morphological properties	Label free biochips	Isolation of CTCs by microfluidics, by exploiting the different physical properties of cancer cells compared with blood cells. Cancer cells are retained in physical structures placed in a microchannel which serve as traps, due to their larger size and lower deformability; successful isolation of CTCs has been demonstrated using cell line colon and breast cancer cells, and blood samples of metastatic lung cancer patients [114]
	Dean flow fractionation	Consists of spiral microchannel with inherent centrifugal forces; enables label-free, size-based isolation of CTCs; diluted whole blood and a sheath fluid are pumped through two different inlets at a certain velocity; cell separation occurs under influence of drag forces; as larger CTCs are exposed to strong inertial lift forces was shown to effectively isolate CTCs from metastatic lung cancer patients’ blood with high sensitivity [115]
Physical and morphological properties (cont.)	ApoStream^™ technology – dielectrophoresis field flow fractionation (DEP-FFF)	Utilizes difference in physical properties of tumor cells, namely the membrane capacitance, for separation. Chromatographic adaptation of microchip dielectrophoresis; DEP can discriminate between cells of similar size with different morphological features in contrast to size-based filtration devices; therefore, isolation of cancer cells from hematologic cells with similar size is possible [116,117]
	Fiber-optic array scanning technology	Automated microscope; allows continuous scanning to enumerate and characterize CTCs; allows detailed cytomorphologic analysis after using an enrichment-free immunofluorescent staining protocol. Comparison of morphologic characteristics of primary and metastatic tumor cells to CTCs from colorectal cancer patients has been performed: showed significant histological congruence, as CTCs represent a random sample of the pleomorphic cells in the primary tumor [118]

CTC: Circulating tumor cell; ISET: Isolation by size of epithelial tumor cells.

Key issues

• • Serum measurement of carcinoembryonic antigen is a proven method of detecting liver metastases from colorectal cancer. It has limited prognostic and predictive value.

• • Tumor-specific markers have demonstrated predictive value that can direct adjuvant chemotherapy. KRAS mutation status predicts poor response to EGFR inhibitors such as cetuximab. Intact SMAD4 expression predicts improved response to 5-fluorouracil-based chemotherapy.

• • Tumor-specific markers also have prognostic value. Colorectal cancers carrying the BRAF mutation have worse prognosis. While loss of heterozygosity at chromosome 18q does not have prognostic significance, loss of SMAD4, which is encoded by 18q, does confer worse prognosis.

• • Circulating tumor cells (CTCs) play a critical role in the process of metastasis. Genetic exchange among tumor cells can lead to the development of highly metastatic phenotypes within the same patient.

• • The heterogeneity of the genetic profiles of cells from the primary tumor, metastatic tumors and CTCs may be an explanation for variable response to EGFR-inhibitor chemotherapy.

• • CTCs are not only a marker for advanced disease, but also have prognostic potential. In addition, CTCs have predictive power, as a decrease in CTC level during chemotherapy is correlated with improved response to chemotherapy.

• • There are numerous methods for the enrichment and detection of CTCs. Further studies are required for harmonization of these methods before CTCs can be adapted for widespread clinical use.

• • Single-cell analysis of CTCs, as well as cells from primary and metastatic tumors, has the potential to enable more targeted therapy for colorectal cancer by identifying different subpopulations of cancer cells responsible for metastatic growth and spread that may require different treatments.

• • Cell-free DNA has the potential to be an easily obtainable, powerful alternative for monitoring therapeutic response in colorectal cancer patients.

• • As for CTCs, there is also a large number of different detection methods for cell-free DNA, which requires standardization and clinical trial validation.

Financial & competing interests disclosure

This article was supported in part by the John and Marva Warnock Cancer Research Fund (SS Jeffrey) and fellowship funds from Natalie and Vladimir Ermakoff (E Kidess). SS Jeffrey is an inventor of the MagSweeper technology discussed in this manuscript. Stanford University has licensed this technology to Illumina, Inc., and receives licensing royalties; SS Jeffrey has donated her royalties to support student programs at The Jackson Laboratory, a nonprofit biomedical research institution. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript.

No writing assistance was utilized in the production of this manuscript.