Characterizing the Cancer Genome in Blood

Abstract

Cell-free circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs) can be found in the bloodstream of individuals with cancer and are increasingly being explored as biomarkers in various aspects of cancer management. The application of next-generation sequencing (NGS) technologies to ctDNA and CTC analysis are providing new opportunities to characterize the cancer genome from a simple blood test and can facilitate the ease with which tumor-specific genomic changes can be followed over time. The serial analysis of ctDNA and CTCs has enormous potential to provide insights into intratumor heterogeneity and clonal evolution during disease progression, and may ultimately allow noninvasive molecular disease monitoring to guide therapeutic decisions and improve patient outcomes.

The advent of next-generation sequencing (NGS) studies have provided important insights into the role of clonal genome evolution in cancer (Aparicio and Caldas 2013). Cancers evolve during disease progression and under the selective pressure of anticancer therapies. The process of clonal evolution is in part driven by intratumoral heterogeneity and is a dynamic process of subclonal expansion and diversification. Today, the inability to accurately characterize spatial and temporal heterogeneity within tumors is thought to be a main reason for the present-day failure of many cancer treatments. Current technological advances in the detection and characterization of circulating biomarkers (circulating tumor DNA [ctDNA] and circulating tumor cells [CTCs]) are providing new opportunities for treatment tailoring that are based on monitoring tumor evolution in real time. Circulating biomarkers can provide a “liquid biopsy” alternative to tissue biopsies allowing noninvasive tumor genotyping and facilitating the serial analysis of genomic changes from a simple blood test. The application of NGS to ctDNA and CTC analysis has marked potential to aid our understanding of cancer heterogeneity, disease evolution, and metastatic biology. Here I coalesce the most significant recent findings in the field and focus on the promises and challenges in developing these tools for clinical applications.

CIRCULATING TUMOR DNA (ctDNA)

The identification of cell-free circulating DNA (cfDNA) in the blood was first described in 1948; however, the benefit of measuring circulating nucleic acids for biomarker applications in cancer has only begun to be thoroughly explored in the last decade (Mandel and Metais 1948). The majority of cfDNA is released from hematopoietic cells in healthy individuals and the genomic and epigenetic features of cfDNA reflect the genome and epigenome of the cell of origin (Sun et al. 2015; Lehmann-Werman et al. 2016; Snyder et al. 2016). In healthy individuals, the quantity of cfDNA is generally very low (<10 ng/mL of plasma) but this can increase by 5–10 times in patients with malignant disease. Greater amounts of cfDNA are identified in cancer patients compared with those found in healthy controls because of the presence of ctDNA that contains tumor-specific sequences harboring somatic genomic alterations found in a patient’s tumor (Anker et al. 1999; Gormally et al. 2007). ctDNA enters the circulation after apoptosis and/or necrosis of tumor cells and is usually fragmented to approximately 160–180 bp, which reflects the degradation of DNA into nucleosomal units, which is characteristic of the apoptotic process (Jahr et al. 2001; Stroun et al. 2001; Mouliere et al. 2011). Whereas ctDNA has the potential to be released into the bloodstream from both tumor tissue and the lysis of CTCs, CTCs and ctDNA represent distinct entities, with the literature now showing that despite the absence of detectable CTCs it is still possible to identify ctDNA (Punnoose et al. 2012; Dawson et al. 2013; Bettegowda et al. 2014).

Measuring absolute levels of cfDNA have been investigated for various applications in cancer management; however, elevated levels of nonspecific cfDNA can also be identified in healthy individuals and patients with benign diseases limiting the potential clinical utility (Gormally et al. 2007). In contrast, monitoring tumor-specific genetic aberrations in the form of ctDNA has excellent specificity and the potential to serve as a highly sensitive biomarker for patients with cancer. ctDNA can be detected in a number of various solid malignancies and the fraction of ctDNA can vary greatly from <0.1% to >50%. Whereas levels are higher in patients with advanced cancers, recent studies have shown that ctDNA can be detected in patients with localized disease, supporting the broad applicability of ctDNA as a biomarker across various tumor types and in different stages of disease (Bettegowda et al. 2014).

METHODOLOGIES FOR ANALYSIS OF ctDNA

The analysis of ctDNA is a challenge requiring extremely sensitive techniques because of the small fraction of tumor-specific DNA present within background levels of normal cfDNA. In early stage disease, the most sensitive analysis techniques are required owing to the small fraction of detectable ctDNA. In patients with advanced malignancies, varied methodological techniques have been successfully used owing to the higher ctDNA fraction, which has allowed more extensive genomic analysis techniques to be applied.

Preanalytical factors can have a notable impact on ctDNA analysis (El Messaoudi et al. 2013). Plasma is favored more than serum and should be processed and stored immediately after blood collection to avoid increases in cfDNA levels because of cell lysis of normal blood cells that might impact relative levels of ctDNA. Using commercially available kits, ctDNA can be extracted from plasma and the analysis can go forward via techniques developed to detect genomic alterations. Progress in genomics technologies is affording excellent opportunities to characterize ctDNA. Types of tumor-specific aberrations that have been followed in plasma include point mutations, chromosomal rearrangements, copy number aberrations, and epigenetic alterations (Fig. 1) (Diehl et al. 2008; Leary et al. 2010, 2012; McBride et al. 2010; Forshew et al. 2012; Chan et al. 2013a,b; Dawson et al. 2013; Murtaza et al. 2013; Carreira et al. 2014; Newman et al. 2014). The types of genomic changes analyzed and the different methodologies used are largely driven by the underlying genomic landscape of each tumor type and the particular clinical application for ctDNA testing.

Figure 1.

Detection of circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs). Cancers release cell-free ctDNA and CTCs into the bloodstream and these can be used as biomarkers to follow various tumor-specific parameters from a simple blood test in cancer patients.

The standard paradigm to discriminate ctDNA from normal cfDNA involves identification of somatic genomic alterations from tumor tissue, the design of assays to recognize these alterations, and the application of these assays to accurately quantify the amount of ctDNA in plasma using polymerase chain reaction (PCR). The detection of somatic point mutations in plasma DNA has been attained via diverse allele-specific PCR and other approaches (Shi et al. 2007; Chen et al. 2009; Board et al. 2010; Sharma et al. 2011; Thierry et al. 2014); however, digital PCR has now emerged as one of the most sensitive analysis tools for ctDNA detection (Vogelstein and Kinzler 1999). Current systematic procedures that involve the use of digital PCR include droplet-based systems (Hindson et al. 2011), microfluidic platforms (Wang et al. 2010), and using beads, emulsions, amplification, and magnetics (BEAMing) (Dressman et al. 2003; Diehl et al. 2008). Digital PCR approaches offer high sensitivity for identifying mutations in ctDNA at low allele fractions (AFs) (∼0.01%). However, they are most relevant when a finite number of genomic loci are evaluated, for example, the identification of predefined or hot-spot mutations in ctDNA. In parallel to the analysis of somatic point mutations, chromosomal rearrangements (e.g., translocations or gains/losses of chromosomal regions) may also be detected in ctDNA using digital PCR (McBride et al. 2010). Personalized analysis of rearrangement ends (PAREs) is an approach that involves identifying specific somatic rearrangements in tumor tissue and the design of PCR-based assays to recognize these in plasma DNA (Leary et al. 2010). These approaches offer high levels of sensitivity (∼0.001%) and inherent specificity for ctDNA detection given that the tumor-specific rearrangements are not present in normal cells. However, as with other single-locus PCR-based assays, this approach relies on the initial identification of rearrangements in tumor tissue and limits the ability to track multiple genomic alterations in parallel.

NGS technologies are being increasingly applied to plasma DNA analysis, providing the opportunity to bypass the need for prior analysis of tumor tissue and allowing more comprehensive analysis across larger genomic areas. Targeted deep sequencing using PCR amplicon-based (e.g., Kinde et al. 2011; Forshew et al. 2012; Carreira et al. 2014; Rothe et al. 2014) or hybrid capture (e.g., Newman et al. 2014, 2016; Lanman et al. 2015; Phallen et al. 2017) target enrichment have been used to sequence specified genomic regions in plasma DNA and detect mutations in panels of commonly altered genes across various tumor types. In addition, whole-exome analysis has also been successfully applied to plasma DNA analysis opening up new opportunities to characterize ctDNA without needing to focus on predefined or existing mutations (Murtaza et al. 2013). Finally, in selected cases, whole-genome sequencing (WGS) of plasma DNA has been possible, providing a novel view of chromosomal and copy number aberrations in ctDNA genome-wide (Leary et al. 2012; Chan et al. 2013b; Heitzer et al. 2013b).

NEXT-GENERATION SEQUENCING OF CIRCULATING TUMOR DNA

Targeted Sequencing

The first NGS approaches to be applied directly to ctDNA analysis have involved the use of targeted deep sequencing to interrogate candidate cancer genes. Target-enrichment strategies have allowed mutations to be identified at low AFs across sizeable genomic regions from a few nanograms of fragmented template from circulating DNA. The most common methods for target enrichment involve (1) highly multiplexed PCR to generate amplicon libraries for sequencing, and (2) hybrid capture using pools of oligonucleotide probes designed to target specific regions of interest from fragment libraries. In contrast to single-locus assays, targeted sequencing approaches allow multiple mutations to be tracked in parallel, providing advantages for tumor monitoring and insights into clonal changes.

Amplicon-based targeted sequencing approaches involve multiplexed PCR for target enrichment followed by deep sequencing, and these approaches have been highly successful for ctDNA analysis across several malignancies (Forshew et al. 2012; Dawson et al. 2013; Murtaza et al. 2013; Carreira et al. 2014; Rothe et al. 2014; Christie et al. 2017; Yeh et al. 2017a,b). Importantly, the success of multiplex PCR strategies for target enrichment relies on the construction of libraries with minimal bias and high levels of molecular complexity. Furthermore, due to the potential for errors to be introduced during the amplification process used for library preparation and the sequencing itself, caution must be observed in calling mutations at low AFs (<1%) to avoid false-positive results. In contrast to multiplex PCR strategies, enrichment of genomic regions for targeted sequencing can also be achieved through hybrid capture using oligonucleotides (“capture probes”) complementary to the gene regions of interest (Newman et al. 2014; Lanman et al. 2015). Regions of interest for targeted sequencing may range from a few exons through to large regions of the genome, such as the whole exome (Murtaza et al. 2013).

Standard targeted panels for gene sequencing can detect mutations with an AF of ∼1%; however, strategies to reduce background error rates (e.g., through molecular barcoding) have the potential to allow ctDNA to be detected at AFs below 0.1%. Molecular barcoding involves the attachment of unique molecular sequences to each fragment when creating a sequencing library so that reads originating from the same molecule can be identified and PCR or sequencing errors can be corrected (Kinde et al. 2011; Newman et al. 2016; Phallen et al. 2017). In contrast to smaller targeted panels, the utility of whole-exome sequencing for mutation detection from plasma DNA is currently limited by sensitivity, making it most applicable to patients with advanced malignancies, in which the median mutation burden in plasma is >5% (Murtaza et al. 2013).

Whole-Genome Sequencing

For a genome-wide view of tumor DNA in the circulation, WGS can provide a comprehensive analysis of plasma DNA (Leary et al. 2012; Chan et al. 2013b). The sensitivity of WGS approaches is dependent on the overall amount of ctDNA and the level of sequence data obtained. Low-depth or shallow WGS (∼0.1× coverage) can be used to detect cancer-specific copy number alterations; however, it requires a ctDNA fraction of at least 5%–10%, providing limited sensitivity for profiling ctDNA in early-stage disease (Heitzer et al. 2013b).

In contrast to the detection of point mutations using the targeted sequencing approaches described above, substantially more sequencing is needed to detect structural alterations and genome-wide copy number alterations from low levels of DNA using WGS. A sensitivity of ctDNA detection of ≤0.1% is considered necessary to detect patients with potentially curative tumors and, in the short term, targeted sequencing approaches are likely to be most feasible in this context. However, with continued improvements in NGS approaches and ongoing reductions in costs, a variety of NGS techniques will no doubt play an increasing role in ctDNA analysis for prospective clinical applications in patients with both early- and advanced-stage malignancies.

CIRCULATING TUMOR CELLS

The development of metastatic disease remains the most significant challenge in the treatment of cancer patients and is responsible for the vast majority of cancer deaths. Metastases begin with the dissemination of single cells (Aceto et al. 2014). Blood represents the hematogenous route for tumor dissemination and studying CTCs offers the promise of providing a better understanding of the biology of metastatic disease. The identifying of CTCs in peripheral blood has been of interest for more than a century when Ashworth (1869) first described a case of cancer in which cells that resembled those in the tumor were seen in blood (Ashworth 1869). CTCs have now been successfully identified in patients before primary tumor detection in early stage cancer and when the carcinoma recurs in the setting of metastatic disease (Allard et al. 2004; Cristofanilli et al. 2004; Almokadem et al. 2005). In some patients, they also appear to persist long after removal of the primary tumor reflecting the possibility of tumor dormancy (Meng et al. 2004). Recent research has focused on the development of improved methods to reproducibly identify tumor cells in the circulation and the application of NGS to CTC molecular analysis.

METHODOLOGIES FOR CTC ISOLATION

CTCs are uncommon events that occur at a frequency of roughly one tumor cell per 1 × 10⁷ peripheral blood mononuclear cells (Ross et al. 1993). New methodology to identify CTCs needs to distinguish between epithelial and hematopoietic cells in blood and a vast array of technologies have now emerged. These methods generally involve the enrichment of CTCs based on the physical properties of the cells or expression of cell-surface markers, combined with specific CTC identification using immunologic, molecular, or functional methods. The method most widely used for CTC enrichment to date has been immunomagnetic separation based on expression of epithelial cell-adhesion molecule (EpCAM). The semiautomated CellSearch system (Janssen Diagnostics, Raritan, NJ) is based on the enumeration of epithelial cells, which are separated from the blood by EpCAM antibody-coated magnetic beads and then identified with the use of DAPI and fluorescently labeled antibodies to cytokeratins (8, 18, and 19) (Allard et al. 2004; Riethdorf et al. 2007). The CellSearch system is the only CTC assay to have been fully validated and it has received approval by the U.S. Food and Drug Administration to guide prognosis in patients with metastatic breast cancer, prostate cancer and colorectal cancer (Cristofanilli et al. 2004; Cohen et al. 2008; De Bono et al. 2008; Bidard et al. 2014).

In addition to the CellSearch system, a large number of new technologies are currently in development. The majority of these continue to be based on EpCAM enrichment technologies (e.g., Talasaz et al. 2009; Saucedo-Zeni et al. 2012; Harb et al. 2013; Ozkumur et al. 2013). However, not all CTCs express EpCAM and at least a subset of CTCs undergo epithelial–mesenchymal transition (EMT) whereby epithelial markers may be lost or down-regulated (Yu et al. 2013). Alternative technologies that use marker-independent enrichment methods are therefore needed (Marrinucci et al. 2012; Sarioglu et al. 2015; Au et al. 2017). Moreover, single-step methods of CTC enrichment do not isolate a pure population of CTCs, and persistent leukocyte contamination poses significant challenges for downstream molecular profiling. Following CTC enrichment, new technologies are now being developed whereby individual CTCs can be captured, visualized, and isolated for future analysis (e.g., Gascoyne et al. 2009; Fabbri et al. 2013; Ozkumur et al. 2013; Peeters et al. 2013). Although new research platforms will require validation to ensure reproducible sensitivity and achieve diagnostic standards, these technology advances are improving the opportunities for genomic and transcriptomic analysis of this rare cell population.

NEXT GENERATION SEQUENCING ANALYSIS OF CTCs

CTC Mutational Profiling

Initial genetic studies involving CTCs were largely performed on DNA extracted from enriched CTC populations, and the presence of wild-type DNA from contaminating leukocytes has posed significant challenges. For example, allele-specific PCR-based assays of CTC-enriched cell populations have been used to detect epidermal growth factor receptor (EGFR) mutations in non-small-cell lung cancer (NSCLC); however, some studies have shown poor concordance between tumor biopsies and CTC-derived genotypes owing to limited sensitivity from the presence of leukocyte-derived DNA (Punnoose et al. 2012). These limitations can be overcome if single CTCs are isolated and subjected to genomic analysis. Single-cell CTC analysis provides the opportunity to assess heterogeneity between CTCs and the ability to identify mutations coexistent within a cell. However, single-cell CTC analysis raises new challenges, as a single cell contains only 6.6 pg of DNA. For NGS to be applied for single-cell analysis, whole-genome amplification is required because of the minimal amount of starting material. The amplification must be uniform and reliable, and caution must be taken to ensure that the amplification does not introduce errors or amplification bias that may confound sequencing results (Dean et al. 2002). Furthermore, stringent bioinformatic approaches are required to ensure reliable identification of tumor-specific changes in single CTCs. Single-cell CTC mutational analysis has now been reported in patients with colorectal, lung, and prostate malignancies (Heitzer et al. 2013b; Ni et al. 2013; Lohr et al. 2014). Although still early in development, the prospect of performing single-cell CTC analysis holds great promise to improve our understanding of the function of individual CTCs and how clonal changes evolve over time.

In parallel to the molecular characterization of individual CTCs, intense interest surrounds the use of isolating viable CTCs and using these as a tissue source for functional assays and drug sensitivity testing. CTCs have been cultured ex vivo in patients with breast cancer and activating ESR1 mutations, known to be associated with hormone resistance, have been identified in CTC-derived cell lines from aromatase inhibitor-pretreated estrogen receptor (ER)-positive metastatic breast cancer patients (Yu et al. 2014). In breast cancer and small-cell lung cancer (SCLC), CTCs have been shown to be tumorigenic in immunocompromised mice, confirming that CTC-enriched populations contain tumor-initiating cells (Baccelli et al. 2013; Hodgkinson et al. 2014). In SCLC, the resultant CTC-derived explants have been shown to mirror the underlying tumor in terms of their genomic landscape and patterns of drug sensitivity (Hodgkinson et al. 2014). Furthermore, single-cell WGS of CTCs compared with CTC-derived explants has revealed considerable similarity in terms of copy number profiles.

CTC Gene Expression Profiling

In parallel to characterizing genomic changes, the ability to assess gene expression in CTCs provides a powerful opportunity to correlate genetic aberrations with specific transcriptional profiles. Early studies of gene expression have used reverse transcription PCR (RT-PCR) to analyze transcripts in CTC-enriched populations. In these studies, digital subtraction of background signal from leukocytes has been essential to derive CTC-based expression signatures. More recently, isolation of single CTCs and analysis of single-cell transcription profiles using NGS technologies has been adopted. In an effort to overcome some of the technical challenges associated with generating expression profiles from single cells, Ramskold et al. (2012) developed a novel messenger RNA sequencing (mRNA-Seq) protocol demonstrating improved read coverage, sensitivity, and reproducibility, compared with standard RNA sequencing. They applied their methodology to the analysis of melanoma CTCs, revealing that highly expressed transcripts were consistently and accurately discovered in the CTCs and that these were similar to the gene expression profiles of melanocytes and melanoma cell lines, yet different in nature from that of leukocytes.

In patients with metastatic prostate cancer, Cann et al. (2012) similarly showed the ability to perform mRNA-Seq on isolated single CTCs. In this study, the RNA from patient CTCs showed signs of significant degradation consistent with reports of apoptosis among CTCs; however, despite this, tumor-specific transcriptional profiles were still readily detectable. RNA sequencing has also been successfully applied to analyze CTCs enriched from patients with pancreatic cancer, single CTCs isolated from prostate cancer patients, and viable CTCs cultured ex vivo from patients with breast cancer (Yu et al. 2012, 2014; Ting et al. 2014; Miyamoto et al. 2015). These early studies showed the feasibility of NGS in CTC analysis and highlighted the potential of using CTC genomic and transcriptomic profiles to improve our understanding of cancer heterogeneity. Currently, the number of CTCs that would need to be analyzed to provide a global representation of the underlying tumor is uncertain and may vary across tumor type. However, these approaches undoubtedly hold great promise to provide important biological insights into the metastatic process.

CLINICAL APPLICATIONS OF ctDNA AND CTC ANALYSIS

The application of NGS approaches to ctDNA and CTC analysis are currently continuing to evolve (Fig. 2). Although the investigation of ctDNA and CTCs present multiple technical hurdles, continuing refinements are expected in the near future, and both approaches hold major promise as biomarkers in cancer management. In time, it is possible that ctDNA and CTCs will have complementary roles, although separate approaches may be advantageous in certain clinical contexts. ctDNA analysis is desirable because of the ease with which plasma can be analyzed without needing to enrich and isolate a rare population of cells. Genomic study of ctDNA can be applied as a high-throughput strategy for analysis of clinical samples, but it is limited in the scrutiny of point mutations, structural rearrangements, copy number alterations, and DNA methylation. Contrastingly, detailed examination of CTCs brings a special opportunity to investigate the whole cell, allowing DNA- and RNA-based molecular profiling, as well as functional studies for modeling personalized treatment selection. The most promising technology platforms to isolate CTCs are currently not widely available, and future standardization and validation of these approaches will be needed for CTC analysis to be more widely adopted for clinical applications.

Figure 2.

Potential clinical applications of ctDNA and CTCs. The analyses of ctDNA and CTCs have potential to be used in multiple facets of cancer management to guide clinical decisions at the time of diagnosis, disease relapse, and throughout disease progression and treatment.

Tumor Genotyping: Tissue versus Liquid Biopsy

Studies characterizing the genomic landscape of various malignancies have emphasized the diversity in cancer genomes and the importance of spatial and temporal intratumor heterogeneity (Gerlinger et al. 2012; Nik-Zainal et al. 2012). The sampling of a single region through a tumor biopsy limits the extent to which the complete spectrum of mutations can be assessed. Moreover, serial sampling of tumor material through repeated biopsies is often not possible and hampers efforts to understand genomic evolution during progression and treatment of disease. Current reports suggest that somatic mutations identified in ctDNA and CTCs are largely representative of the underlying tumor genome, providing an alternate noninvasive procedure of tumor sampling that may overcome these limitations (Chan et al. 2013b; Heitzer et al. 2013a; Murtaza et al. 2013; Ni et al. 2013; De Mattos-Arruda et al. 2014; Hodgkinson et al. 2014; Lohr et al. 2014; Rothe et al. 2014; Savas et al. 2016; Yeh et al. 2017a).

An immediate clinical application of this tactic is in identifying specific genomic alterations to steer selected therapies (e.g., EGFR mutations and EML4-ALK rearrangements in NSCLC, BRAF mutations in melanoma, KRAS mutations in colorectal cancer, and HER2 amplification and PIK3CA mutations in breast cancer) (Higgins et al. 2012; Misale et al. 2012; Oxnard et al. 2016; Santiago-Walker et al. 2016). The ability to perform blood-based tumor genotyping assays from ctDNA or CTCs will facilitate which therapeutic targets can be identified in patients and guide treatment decisions, and, importantly, also permit real-time monitoring during disease progression and therapy (Schwaederle et al. 2016a,b). With development of future genotype-driven therapies, clinical applications in this area will most likely increase in the coming years.

Monitoring Tumor Burden and Therapeutic Responses

CTC numbers as assessed by the CellSearch system have been used to provide information on tumor burden and treatment response in several solid malignancies. CTC numbers are highly prognostic in patients with metastatic breast (Cristofanilli et al. 2004), colorectal (Cohen et al. 2008), and prostate cancer (De Bono et al. 2008), and, when followed longitudinally, changes in CTC numbers have also been shown to be a marker of treatment response (Bidard et al. 2014). However, CTC enumeration alone provides no additional information on the molecular profile of the cells and the application of NGS to CTC analysis will provide improved opportunities for molecular disease monitoring in the future.

Recent studies have examined ctDNA dynamics in an analogous fashion, investigating the relationship between ctDNA levels, tumor burden, and treatment response (Diehl et al. 2008; Yung et al. 2009; Leary et al. 2010; McBride et al. 2010; Dawson et al. 2013; Bettegowda et al. 2014; Gray et al. 2015; Roschewski et al. 2015). Levels of ctDNA in plasma closely correlate with tumor size and disease stage (Bettegowda et al. 2014). Following treatment, dynamic changes in ctDNA levels reveal changes in tumor burden, and increases in ctDNA often predate detection of progressive disease using radiological methods by several months (Diaz et al. 2012; Dawson et al. 2013; Parkinson et al. 2016). Quantitative analysis of ctDNA may prove to be an important measure of outcome. Initial data has reinforced an association between ctDNA levels and prognosis in patients who have advanced disease (Dawson et al. 2013; Nygaard et al. 2013; Santiago-Walker et al. 2016), but future studies will be required to support the role of ctDNA as a surrogate biomarker for disease-free and overall survival in larger patient populations.

A significant hurdle cancer patients face is the development of resistance to chemotherapeutic and targeted agents. This represents another area in which ctDNA and CTC analyses may play a key role. Resistance is in part developed because of an evolving spectrum of somatic mutations within the tumor under the selective pressure of treatment. Current research has shown that ctDNA and CTC analysis can detect emerging mutations associated with treatment resistance, which includes the identification of T790M-EGFR mutations in NSCLC (Maheswaran et al. 2008; Murtaza et al. 2013; Oxnard et al. 2014; Chabon et al. 2016), KRAS mutations and mesenchymal-epithelial transition factor (MET) amplification in colorectal cancer patients receiving EGFR-based therapies (Diaz et al. 2012; Misale et al. 2012, 2014; Bardelli et al. 2013; Siravegna et al. 2015), the development of ESR1 mutations in hormone-resistant breast cancer (Yu et al. 2014; Schiavon et al. 2015), and the development of androgen receptor mutations and copy number changes, as well as the androgen receptor splice variant 7 (ARV7) in patients receiving treatment for advanced prostate cancer (Carreira et al. 2014; Romanel et al. 2015; Scher et al. 2016). The application of serial ctDNA analysis through NGS of plasma DNA has also been shown to provide a comprehensive and unbiased assessment of genomic changes and tumor evolution during the acquisition of treatment resistance (Murtaza et al. 2013, 2015; Chabon et al. 2016; Scherer et al. 2016; Abbosh et al. 2017; Yeh et al. 2017b). Furthermore, the ability to isolate viable CTCs, culture CTCs ex vivo, and generate CTC-derived xenograft mouse models is opening up a vast array of new opportunities to perform drug sensitivity testing and study drug resistance mechanisms to guide personalized treatment approaches (Hodgkinson et al. 2014; Yu et al. 2014).

Early Detection and Monitoring Minimal Residual Disease

In hematological malignancies, assessing recurrent genomic alterations such as the BCR-ABL and PML-RARα translocations, are routinely used as biomarkers to monitor minimal residual disease (Bregni et al. 1989; Hughes et al. 2006). For solid malignancies, tumor-specific genomic alterations identified on an individual basis could be used to guide clinical management in an analogous fashion. ctDNA and CTC analyses have the possibility to be useful as biomarkers following curative treatment to identify patients who may be at risk of relapsing. Earlier literature has shown that monitoring tumor-specific mutations in plasma following surgical resection, can allow individuals with residual disease to be identified and preemptively detect disease recurrence (Diehl et al. 2008; Chen et al. 2009; Garcia-Murillas et al. 2015; Olsson et al. 2015; Reinert et al. 2016; Tie et al. 2016). Early diagnosis of relapse, while disease burden is minimal, may allow the introduction of additional effective treatments. Conversely, ctDNA or CTCs may assist in stratifying those patients at highest risk of relapse to guide the most appropriate selection of adjuvant therapy. Following treatment for early-stage malignancies, additional research will be necessary to characterize the relationship between ctDNA, CTCs, and disease outcomes.

Last, the potential use of ctDNA or CTCs as biomarkers for cancer screening is one of the most challenging future applications, requiring the highest levels of sensitivity, specificity, and reproducibility to achieve this goal. Ongoing technological advancements may lead to the application of these approaches as screening tools to allow early identification of malignancy at a stage when curative treatments can be offered (Gormally et al. 2006; Beaver et al. 2014; Cohen et al. 2016; Phallen et al. 2017).

CONCLUDING REMARKS

Cancers evolve during disease progression and under the selective pressure of therapy and our ability to continually monitor tumor-related changes over time will be critical in future cancer management to select optimal treatment regimens. The application of NGS to ctDNA and CTC analysis is providing a one-of-a-kind opportunity to facilitate personalized treatment decisions by allowing patient-specific genomic changes to be monitored in real time via a minimally invasive technique. Although circulating biomarkers hold great promise in cancer management, substantial effort is still required to understand their clinical application in various contexts. Future research needs to focus on optimizing and standardizing NGS technologies for both ctDNA and CTC analysis and establishing the clinical utility of ctDNA and CTC testing through appropriately designed prospective clinical trials.