Clinical Utility of Analyzing Circulating Tumor DNA in Patients with Metastatic Colorectal Cancer

This review focuses on the development of circulating tumor DNA (ctDNA) analyses for detecting predictive biomarkers for metastatic colorectal cancer and highlights prospective clinical trials to demonstrate the usefulness of ctDNA analysis.

Abstract

Multiple genomic changes caused by clonal evolution induced by therapeutic pressure and corresponding intratumoral heterogeneity have posed great challenges for personalized therapy against metastatic colorectal cancer (mCRC) in the past decade. Liquid biopsy has emerged as an excellent molecular diagnostic tool for assessing predominant spatial and temporal intratumoral heterogeneity with minimal invasiveness.

Previous studies have revealed that genomic alterations in RAS, BRAF, ERBB2, and MET, as well as other cancer‐related genes associated with resistance to anti‐epidermal growth factor receptor (EGFR) therapy, can be analyzed with high diagnostic accuracy by circulating tumor DNA (ctDNA) analysis. Furthermore, by longitudinally monitoring ctDNAs during anti‐EGFR therapy, the emergence of genomic alterations can be detected as acquired resistance mechanisms in specific genes, mainly those associated with the mitogen‐activated protein kinase signaling pathway. Analysis of ctDNA can also identify predictive biomarkers to immune checkpoint inhibitors, such as mutations in mismatch repair genes, microsatellite instability‐high phenotype, and tumor mutation burden. Some prospective clinical trials evaluating targeted agents for genomic alterations in ctDNA or exploring resistance biomarkers by monitoring of ctDNA are ongoing.

To determine the value of ctDNA analysis for decision‐making by more accurate molecular marker‐based selection of patients and identification of resistance mechanisms to targeted therapies or sensitive biomarkers for immune checkpoint inhibitors, clinical trials must be refined to evaluate the efficacy of study treatment in patients with targetable genomic alterations confirmed by ctDNA analysis, and resistance biomarkers should be explored by monitoring ctDNA in large‐scale clinical trials. In the near future, ctDNA analysis will play an important role in precision medicine for mCRC.

Implications for Practice.

Treatment strategies for metastatic colorectal cancer (mCRC) are determined according to the molecular profile, which is confirmed by analyzing tumor tissue. Analysis of circulating tumor DNA (ctDNA) may overcome the limitations of tissue‐based analysis by capturing spatial and temporal intratumoral heterogeneity of mCRC. Clinical trials must be refined to test the value of ctDNA analysis in patient selection and identification of biomarkers. This review describes ctDNA analysis, which will have an important role in precision medicine for mCRC.

Introduction

Over the last 20 years, the clinical outcome of patients with metastatic colorectal cancer (mCRC) has improved because of the increased number of patients who undergo surgical resection of metastatic lesions and the development of systemic therapies, including targeted agents [1]. However, the prognosis for patients with mCRC remains poor, with a median overall survival of approximately 30 months. Although vast efforts and investments have been made in the discovery of novel biomarkers to predict the efficacy of anticancer agents for mCRC, extended RAS, BRAF, and microsatellite instability (MSI) are the only validated predictive biomarkers so far. Further studies are critical to the expansion of the precision medicine armamentarium for patients with mCRC.

Comprehensive molecular analyses of CRC have revealed somatic alterations in multiple genes encoding targetable proteins, including KRAS, NRAS, BRAF, IGF2, IGFR, HER2, HER3, MET, MEK, PIK3CA, AKT, and MTOR [2]. Preclinical and clinical studies are being conducted to determine the role of these molecular abnormalities as predictive and prognostic biomarkers. In addition to identifying potential biomarkers, molecular testing is evolving from the “one test‐one drug” paradigm to multiplex approaches that explore the genomic landscape of a tumor. The evolution of next‐generation sequencing (NGS) technologies has enabled the genotyping of tissue samples for multiple genomic alterations in parallel.

However, tissue‐based NGS has inherent limitations. For example, recent studies revealed that multiple genetic changes can arise because of clonal evolution under treatment pressure in mCRC, resulting in acquired secondary resistance [3]. Therefore, longitudinal surveillance of clonal evolution is required to identify secondary resistance. However, repeated tissue biopsies are expensive and can be difficult to perform because of the inherent risk of complications. Moreover, biopsies or tissue sections represent a single snapshot of the tumor in time and space; therefore, they often fail to detect intratumoral heterogeneity, which is a great challenge for the optimal treatment selection for mCRC [4], [5], [6], [7].

Recent technical advances have enabled analysis of tumor materials obtained from blood or other body fluids. These liquid biopsies can be used to assess the predominant type from intratumoral heterogeneity and overcome the limitations of tissue analyses. Specifically, analysis of circulating tumor DNA (ctDNA), which is tumor‐derived fragmented DNA released into the blood stream, has been shown to have clinical utility in detecting genetic alterations in various types of cancers, including CRC. A number of studies since the early 2000s have revealed that mutations and/or methylations of ctDNA have independent prognostic value for stage I–III CRC [8]. In patients with mCRC, ctDNA analysis potentially evaluates the disease burden of tumors in mCRC by quantification [9]; this can be used with serial monitoring of ctDNA to predict tumor response during treatment. Indeed, some studies have indicated an association between the decrease of ctDNA levels or mutant allele frequency (MAF) during treatment and tumor response [9], [10], [11], [12], [13], [14]. Furthermore, evolving technologies such as digital polymerase chain reaction (PCR) and NGS have expanded the potential for ctDNA analysis, and studies have employed these techniques for targeted therapy of mCRC. In this review, we focus on the development of ctDNA analyses for detecting predictive biomarkers for mCRC and highlight our prospective clinical trials to demonstrate the usefulness of ctDNA analysis.

Identification of Biomarkers Related to Resistance to Epidermal Growth Factor Receptor Blockade in mCRC by ctDNA Analyses

The clinical durability of the treatment of mCRC is limited by primary and secondary resistance. Resistant mechanisms in RAS wild‐type patients with mCRC treated with anti‐epidermal growth factor receptor (EGFR) therapy mainly involve alterations of specific genes associated with the mitogen‐activated protein kinase (MAPK) signaling pathway [15], [16], [17], [18]. Liquid biopsies can detect these genomic alterations related to primary or acquired resistance to anti‐EGFR therapy in ctDNA. Here, we summarize the utility of ctDNA analysis in detecting these biomarkers of anti‐EGFR therapy resistance in mCRC.

RAS Mutations

Concordance of RAS Status Between Tissue and ctDNA Analysis.

Mutations in extended RAS, that is, KRAS exon 2, 3, or 4 or NRAS exon 2, 3, or 4 are negative predictive biomarkers that have been validated in prospective‐retrospective or retrospective analyses in several randomized studies with anti‐EGFR antibodies for mCRC [19], [20], [21], [22], [23], [24]. The utility of detecting RAS mutations by ctDNA analysis has been assessed and validated in several studies with novel liquid biopsy technologies [25], [26], [27], [28], [29], [30], [31] (Table 1). The first blinded prospective study that evaluated KRAS mutation status in ctDNA using Intplex, a quantitative PCR‐based method using originally designed specific primers for multiple gene mutations [32], in mCRC showed high concordance (94%) and specificity (98%) compared with tissue analyses [26]. The OncoBEAM RAS CRC assay is now available as the only test for detecting RAS mutations in ctDNA with European Conformity in vitro diagnostic (CE‐IVD) study; it can accurately detect up to 34 mutations in exons 2, 3, and 4 of KRAS and NRAS genes, using the BEAMing technology [28], [29]. One of the largest cohort prospective studies revealed that the accuracy raised up to 95.6% in patients with liver metastases [31]. Of note, blood samples were obtained before the start of anti‐EGFR therapy in all these validation studies to avoid detecting acquired RAS mutations as discussed below. In addition, the turnaround time was shown to be 18 days and 7 days for tumor tissue and ctDNA analysis in this study, respectively. This result is consistent with findings from studies evaluating the turnaround time of ctDNA analysis of other cancer types [33], [34], [35] and suggests that ctDNA analysis can help patients with mCRC to quickly receive the optimal targeted therapy.

Table 1. Concordance of RAS mutation statuses between tumor‐tissue analysis and ctDNA analysis.

^a74% for multiplex dPCR.

Abbreviations: dPCR, digital polymerase chain reaction; MAF, mutant allele frequency; NA, not available; NE, not evaluated; NGS, next‐generation sequencing; NPA, negative percent agreement; PPA, positive percent agreement.

Predictive Value of Mutant RAS in ctDNA for Efficacy of Anti‐EGFR Therapy.

Comparisons of the progression‐free survival (PFS) of patients with mCRC treated with anti‐EGFR therapy using the tissue RAS versus the plasma ctDNA RAS result to determine eligibility of patients for targeted therapy indicated similar PFS following first‐line [29] and second/third‐line treatments [30]. These initial observations are encouraging and suggest that plasma RAS testing could accurately determine the eligibility of RAS wild‐type patients for anti‐EGFR therapy.

However, the application of plasma RAS testing remains an important challenge to develop clinically meaningful thresholds for the RAS MAF in ctDNA for appropriately selecting patients who may benefit from anti‐EGFR therapy. When a low sensitivity threshold of MAF is applied, ctDNA analysis can identify patients with a very low number of RAS mutant cells who could benefit from anti‐EGFR therapy [36]. Indeed, a prospective‐retrospective study showed that a RAS mutation with a MAF detected by ctDNA ≥0.1 was significantly associated with short PFS after anti‐EGFR therapy, whereas the PFS of patients with a RAS mutation with a MAF detected by ctDNA <0.1 was similar to that of the RAS wild‐type [30]. Further investigations are needed to evaluate the efficacy of anti‐EGFR therapy for mCRC with any mutant RAS with low MAF detected by ctDNA harboring wild‐type RAS based on tissue analysis.

Monitoring RAS Mutation by ctDNA Analysis During Anti‐EGFR Therapy.

RAS mutant clones have been identified as drivers of acquired resistance to anti‐EGFR therapy in clinical and preclinical studies [16], [17], [37], [38], [39]. Acquired KRAS mutations have been suggested to emerge not only from the selection of pre‐existing KRAS mutant subclones but also as a result of ongoing mutagenesis in the cancer during anti‐EGFR therapy [17]. Some studies identified RAS mutations in ctDNA after anti‐EGFR therapy. A case series indicated that detection of RAS mutations in the plasma during anti‐EGFR therapy provided early warning of impending resistance, which was confirmed several months later by imaging [40]. A retrospective analysis of ctDNA from mCRC patients refractory to anti‐EGFR therapy showed that KRAS codon 61 and 146 mutations were more common than in treatment‐naïve patients and the frequency of acquired KRAS mutations was inversely correlated with the time since last anti‐EGFR therapy [41].

A retrospective analysis of ctDNA from mCRC patients refractory to anti‐EGFR therapy showed that KRAS codon 61 and 146 mutations were more common than in treatment‐naïve patients and the frequency of acquired KRAS mutations was inversely correlated with the time since last anti‐EGFR therapy.

Furthermore, monitoring ctDNA during anti‐EGFR therapy indicated the emergence of acquired RAS mutations as well as alterations in other genes, including MET, ERBB2, FLT3, EGFR, and MEK [42]. Interestingly, the temporal withdrawal of an anti‐EGFR antibody resulted in a decline in KRAS‐mutant alleles, suggesting that sensitivity to anti‐EGFR antibody was restored. These findings warrant prospective studies evaluating the strategy of monitoring genomic alterations, including RAS mutations, by ctDNA analyses and temporarily stopping or rechallenging anti‐EGFR therapy.

BRAF Mutations

BRAF V600E mutation is a strong biomarker for poor prognosis of CRC [43]. Accumulating evidence suggests that the BRAF V600E mutation decreases the likelihood of a response to anti‐EGFR monotherapy [18], [44], [45]. Preclinical studies demonstrated that patients with BRAF V600E‐mutant CRC were highly sensitive to dual inhibition of EGFR and the MAPK signaling pathway [46], [47]; these results have prompted additional clinical studies to evaluate the efficacy of combined inhibition of BRAF and MEK, EGFR, or PI3K in patients with BRAF V600E‐mutant mCRC.

Analysis of BRAF V600E mutation in ctDNA by digital PCR has also been validated; this method was shown to have a high accuracy of nearly 100% for detecting BRAF V600E mutations [26]. In a phase Ib study, reductions in BRAF V600E ctDNA allele fraction predicted improved radiographic responses to vemurafenib, a BRAF inhibitor, cetuximab, and irinotecan [48]. In addition, serial NGS‐based ctDNA analysis identified mutations in MAPK signaling pathway‐related genes (MEK1, ARAF, ERBB2, and EGFR) upon progression. In another phase I/II study of the clinical activity of dabrafenib, a BRAF inhibitor, trametinib, a MEK inhibitor, and panitumumab against BRAF V600E‐mutant CRC, the BRAF V600E‐mutant fraction burden in ctDNA was more markedly reduced by week 4 in responders than in nonresponders and the emergence of RAS mutations in ctDNA was detected in 9 of 22 patients (41%) during disease progression [49].

EGFR Mutations

Somatic mutations in EGFR have been demonstrated as a mechanism of resistance to anti‐EGFR therapy. One study identified a point mutation (S492R) within the EGFR ectodomain (ECD) that arose in a cetuximab‐resistant cell line and demonstrated that panitumumab could block EGFR activation in these mutants [50]. Furthermore, the researchers identified the EGFR S492R mutation in a patient with mCRC after treatment with cetuximab, and panitumumab treatment markedly reduced the size of all lesions in this patient. Another study revealed that several other EGFR ECD mutations such as R451C, K467T, S464L, G465R, and I491M were acquired after treatment with cetuximab; these mutants showed different binding properties from cetuximab and panitumumab [51]. Sym004, a 1:1 mixture of two antibodies against nonoverlapping epitopes of EGFR ECD, has been developed to overcome the resistance to anti‐EGFR therapy and demonstrated tumor shrinkage in patients whose disease progressed after anti‐EGFR therapy, including a patient with EGFR S492R mutation in a phase I trial [52]. However, in a randomized phase II study of Sym004, this agent seemed to be more active in patients with no mutant allele of RAS, BRAF, and EGFR ECD in ctDNA compared with investigator's choice [53]. This result indicates the further need for the development of agents to overcome EGFR ECD mutations.

An analysis of ctDNA using the BEAMing method detected EGFR S492R mutation in 5 of 62 patients (8%) who had acquired resistance to anti‐EGFR antibody after an initial response [41]. Notably, four of five patients with ctDNA EGFR S492R mutation also had newly detected KRAS mutations, suggesting that these mutations were not mutually exclusive. Another analysis of ctDNA revealed that EGFR ECD mutations arose in patients who experienced greater and longer responses to anti‐EGFR therapy compared with patients with acquired RAS mutations [54]. This finding indicates that EGFR ECD mutations are acquired as de novo mutations after anti‐EGFR therapy rather than occurring from pre‐existing low levels of mutations.

ERBB2 (HER2) Amplification

ERBB2 amplifications have been observed in approximately 3% of patients with mCRC [55]. Preclinical studies using a patient‐derived xenograft model of ERBB2‐amplified CRC showed that ERBB2‐amplified CRC was resistant to anti‐EGFR treatment and responded to HER2‐targeted therapy [56], [57]. Consistent with these findings, the frequency of ERBB2 amplifications increase from approximately 3% in treatment‐naïve patients to over 10% in patients administered anti‐EGFR treatment [56]. Clinical studies have investigated the efficacy of combined HER2‐targeted therapies, which showed some clinical benefit [58], [59].

To detect copy number alterations (CNAs), including ERBB2 amplification, in ctDNA, researchers have utilized whole‐genome sequencing methods to evaluate the normalized copy number of the target locus in circulating cell‐free DNA [60]; however, these methods have limited sensitivity because of their low depth. Recently, targeted sequencing approaches have been established to detect CNAs in ctDNA [61]. Guardant360, an NGS‐based ctDNA analysis method that targets cancer‐related genes, was shown to identify CNAs for multiple genes, including ERBB2, EGFR, and MET, with high diagnostic accuracy [62]. In a National Surgical Adjuvant Breast and Bowel Project study, acquired ERBB2 amplifications were detected using this technology in two of eight patients with mCRC, who initially had HER2‐negative tumors and showed progression after anti‐EGFR therapy [63]. Digital PCR methods have also been developed to detect ERBB2 amplifications in ctDNA [64], [65]. We have now conducted the TRIUMPH study, an investigator‐initiated, multicenter, phase II study, to evaluate the efficacy and safety of combination therapy with trastuzumab and pertuzumab in patients with HER2‐positive mCRC [66]. This study enrolls patients with RAS wild‐type in tissue and not only HER2‐positive detected by tissue analysis, but also ERBB2‐amplified and RAS wild‐type tissues confirmed by ctDNA analysis using an NGS‐based method, which has a potential benefit to detect acquired ERBB2 amplification with minimal invasiveness. In addition, monitoring using ctDNA analysis during treatment or after disease progression is conducted to explore the resistance mechanisms to HER2‐targeted therapy in this study.

MET Amplification

MET amplification is also responsible for de novo and acquired resistance to anti‐EGFR therapy for mCRC [67]. Cotreatment with MET inhibitors and cetuximab caused robust and long‐lasting tumor shrinkage in patient‐derived xenograft models of MET‐amplified mCRC. In a study investigating the frequency of MET amplification in a large cohort of patients with mCRC, MET amplifications were detected in 1.7% (10/590) of tumor tissue biopsies tested by both fluorescence in situ hybridization and sequencing in the pretreatment cohort [68]. In contrast, an NGS panel detected MET amplification in the ctDNAs of 22.6% (12/53) of patients who had been treated with anti‐EGFR therapy and showed disease progression, which was not detected in patients with pre‐cetuximab treatment. In addition, this ctDNA NGS method did not detect MET amplification in post‐cetuximab patients with RAS mutations, suggesting that MET amplification is a resistance mechanism exclusive from RAS mutation in anti‐EGFR‐treated patients. Furthermore, novel technologies have been developed to evaluate MET amplification with improved analytical sensitivity, using amplification‐associated rearrangements along with normalized counting‐based approaches [69]. In a phase Ib study of the efficacy of cabozantinib, a multiple receptor kinase inhibitor, and panitumumab in patients with RAS wild‐type mCRC, the combination markedly reduced the tumor volume in a patient in whom MET amplification was detected in ctDNA but not in tissue [70].

ctDNA Analysis for Immune Checkpoint Inhibitors in mCRC

The emergence of immune checkpoint inhibitors targeting Cytotoxic T‐lymphocyte associated protein‐4 and programmed death‐1 (PD‐1)/programmed death‐ligand 1 (PD‐L1) has facilitated breakthrough developments in cancer immunotherapy; however, most patients with mCRC did not exhibit significant responses to these agents [71], [72], and thus it is necessary to identify novel biomarkers and combination strategies for mCRC.

It has been hypothesized that deficient mismatch repair (dMMR) or mutations inactivating DNA repair genes, such as POLD, POLE, and MYH, could elicit sensitivity toward immune checkpoint inhibitors because of the high tumor neoantigen load caused by hypermutation of the genome. This hypothesis was supported by the results of a clinical trial with pembrolizumab, an anti‐PD‐1 antibody, which indicated an objective response rate of approximately 40% in dMMR mCRC [73]. Another checkpoint inhibitor, nivolumab, also showed that approximately 30% of patients responded to dMMR/MSI‐high (MSI‐H) mCRC in a phase II study [74]. Most recently, the U.S. Food and Drug Administration approved pembrolizumab for dMMR/MSI‐H cancers and nivolumab for dMMR/MSI‐H mCRC. International phase III trials are now ongoing to evaluate the efficacy of immune checkpoint inhibitors for dMMR mCRC.

ctDNA analyses can serve as a potential tool for exploring predictive biomarkers of immune checkpoint inhibitors. For example, mutations in MMR genes can be detected by ctDNA analyses along with the MSI‐H phenotype using five mononucleotide markers (BAT‐25, BAT‐26, MONO‐27, NR‐21, and NR‐24) and has been validated for clinical use by Personal Genome Diagnostics using PlasmaSELECT. Tumor mutation burden (TMB) that often co‐occurs with dMMR/MSI‐H is another predictive biomarker but is at times difficult to assess with tumor tissue from patients diagnosed in the metastatic setting. NGS‐based ctDNA analysis may estimate blood TMB that reflects neo‐antigenicity at the time when the blood sample is obtained. Khagi et al. reported that PFS after immune checkpoint inhibitors was significantly better in patients with >3 variant of undetermined significance (VUS) alterations versus ≤3 VUS alterations detected by an NGS‐based ctDNA analysis in a population including patients with mCRC [75]. Some resistance biomarkers of somatic genomic alterations, such as JAK1/2 mutation, EGFR mutation, and MDM2 amplification, may be related to the resistance to immune checkpoint inhibitors [76], [77]. Such genomic alterations can also be detected by NGS‐based ctDNA analyses.

ctDNA Analysis‐Adapted Prospective Studies

Clinical Trial Evaluating the Clinical Utility of ctDNA‐Guided Decision‐Making

The eligibility criteria in most clinical trials only includes patients with certain genomic alterations based on analysis of tumor tissue. However, the analysis of archival tissue may not accurately reflect clonal evolution under prior therapeutic pressure, and re‐biopsy samples are difficult to obtain from all patients because of the inherent risk of complications. In contrast, liquid biopsy, which is minimally invasive, allows for patient stratification according to the current status of genomic alterations from plasma obtained immediately before enrollment, although the concordance of the broad range of genomic alteration statuses between newly obtained tumor‐tissue analysis and ctDNA analysis remains unclear. To establish the potential value of ctDNA analysis‐guided decision‐making, clinical trials must be refined to test the efficacy of study treatment in patients with targetable genomic alterations confirmed by ctDNA analysis but not analyzing archival tissue. A few studies in which patient selection is based on ctDNA analysis are underway (Table 2).

Table 2. List of clinical trials incorporating ctDNA analysis for patient selection.

Data mainly gathered from Clinicaltrials.gov up to November 2017. Queried terms were “colorectal cancer” and “liquid biopsy” or “circulating DNA,” and clinical trials in which liquid biopsy was conducted as an inclusion biomarker for patients with metastatic colorectal cancer were selected. Trials with “unknown status” or “completed” were excluded. The TRIUMPH study was added to the list.

Abbreviations: ctDNA, circulating tumor DNA; NR, not reported.

The potential advantage of ctDNA analysis can lead to next‐generation clinical trials, such as basket and umbrella trials based on genomic alterations that reflect spatial and temporal intratumoral heterogeneity. Kim et al. have already reported that a clinical trial using ctDNA‐guided matched therapy showed clinical responses, which was similar to tissue‐based targeted therapy, for patients with non‐small cell lung cancer and gastric cancer [78]. The Targeted Agent and Profiling Utilization Registry, an American Society of Clinical Oncology‐sponsored large basket/umbrella trial, accepts patient selection by using ctDNA analysis (NCT02693535). We have planned an umbrella project for mCRC based on the molecular profile based on ctDNA analysis (Fig. 1). Around 1,000 patients with mCRC whose disease progressed after anticancer therapy will be screened by using an NGS‐based ctDNA analysis in this project (UMIN000029315). The patients who have alterations of ERBB2, BRAF V600E, BRAF non‐V600E, or MET or high tumor mutation burden in tumor tissue or ctDNA will be enrolled into each substudy targeting the genomic alteration. In the substudies, the outcomes will be compared between both patient groups with the genomic alteration based on tumor tissue and ctDNA analysis to assess the potential of ctDNA‐guided biomarker selection compared with the standard‐of‐care tissue analysis. Furthermore, natural history will be followed as a historical control of patients with rare genomic alterations targeted in each substudy and compared with the treatment group.

Figure 1.

Schema of a planning umbrella project for metastatic colorectal cancer (mCRC). Patients with mCRC who have disease progression after anticancer therapy will be screened by ctDNA analysis. Trastuzumab and pertuzumab, a mitotic inhibitor, a triplet of anti‐EGFR and BRAF inhibition and MEK inhibition, a doublet of anti‐EGFR and MET inhibition, and anti‐PD‐1 will be tested for each targetable alteration.

Abbreviations: ctDNA, circulating tumor DNA; EGFR, epidermal growth factor receptor; MT, mutant; PD‐1, programmed death‐1; TMB, tumor mutation burden.

Exploratory Study Using ctDNA Analysis in Prospective Clinical Trials

In addition to the potential of appropriate molecular selection of patients, ctDNA analysis can explore the mechanisms of resistance to targeted therapies by monitoring genomic alterations during treatment. Recently, baseline blood samples were prospectively collected and analyzed in large‐scale clinical trials to identify novel biomarkers. These exploratory analyses with large‐scale trials have the advantage of being able to analyze novel biomarkers in similar populations as intention‐to‐treat populations because of the ease of collecting samples. In the CORRECT trial, which demonstrated the survival benefit of regorafenib in salvage line for mCRC, mutations in KRAS, BRAF, and PIK3CA in ctDNA obtained from the plasma of enrolled patients were analyzed by BEAMing technology [79]. BEAMing analysis of plasma revealed consistent trends toward a clinical benefit with regorafenib irrespective of KRAS and PIK3CA mutational status in ctDNA. Interestingly, although the clinical benefit of regorafenib was not significantly affected by circulating DNA concentrations in the plasma, a high baseline circulating DNA concentration was associated with poor prognosis in the placebo group. In the ASPECCT study, a phase III study of panitumumab versus cetuximab in patients previously treated with KRAS exon 2‐wild‐type mCRC, ctDNAs were analyzed by droplet digital PCR (ddPCR) and NGS. Consistent with the difference in binding properties between the anti‐EGFR antibodies, ctDNA analysis by ddPCR revealed that cetuximab treatment resulted in more frequent emergence of EGFR S492R mutations in the plasma than panitumumab treatment [80]. NGS‐based monitoring of the mutational landscape in ctDNAs of 238 patients receiving panitumumab showed that at least 10% of the subjects acquired mutations during therapy, both inside and outside of the EGFR pathway [81]. The PARADIGM study is a large phase III study that assessed modified FOLFOX6 plus panitumumab or bevacizumab as the first‐line treatment for RAS wild‐type mCRC. Archival formalin‐fixed paraffin‐embedded tumor tissues and ctDNA samples, which are obtained prior to treatment and at the confirmed disease progression stage from similar population as intention‐to‐treat population, will be analyzed to elucidate mechanisms of primary and secondary resistance to anti‐EGFR and anti‐vascular endothelial growth factor therapy in the first‐line treatment [82].

NGS‐based monitoring of the mutational landscape in ctDNAs of 238 patients receiving panitumumab showed that at least 10% of the subjects acquired mutations during therapy, both inside and outside of the EGFR pathway.

Conclusion

Analysis of ctDNA has emerged as a potential tool for evaluating the complicated biological processes involved in CRC, represented by intratumoral heterogeneity and clonal evolution, given the limited availability of precision medicine for patients with mCRC. The practical advantages, such as minimal invasiveness, rapid turnaround time, and low cost of sampling procedures, enable medical oncologists to promptly guide patients to an optimal therapy and potentially monitor the resistance using multiple sampling to implement changes in the therapy. As reviewed above, although several studies have already suggested the utility of ctDNA analysis for evaluating the spatial and temporal dynamics of tumor clones under therapeutic pressures in mCRC, some limitations have been observed. One major limitation of ctDNA analysis is the sensitivity and specificity versus those of tissue analysis. Some studies identified factors of discordance, such as the site of metastasis (peritoneum and lung), histology (mucinous), administration of treatment prior to blood collection, low tumor burden, absence of a primary tumor, and lower levels of lactate dehydrogenase, alkaline phosphatase, and CA19‐9 [29], [30], [31]. Additionally, ctDNA analysis cannot analyze biomarkers other than genomic alterations, such as microRNAs. Analysis of circulating RNAs, circulating tumor cells, and exosomes may overcome this issue.

Until now, most studies included small cohorts of patients, and thus the value and limitations of ctDNA analysis in clinical practice for mCRC must be validated in well‐designed or large‐scale prospective trials. In conclusion, liquid biopsy analysis of ctDNA is creating a paradigm shift in the molecular diagnosis of mCRC, and researchers should design prospective trials to establish ctDNA analysis‐based personalized therapy to improve patient outcomes.

Author Contributions

Conception/design: Yoshiaki Nakamura, Takayuki Yoshino

Collection and/or assembly of data: Yoshiaki Nakamura, Takayuki Yoshino

Data analysis and interpretation: Yoshiaki Nakamura, Takayuki Yoshino

Manuscript writing: Yoshiaki Nakamura, Takayuki Yoshino

Final approval of manuscript: Yoshiaki Nakamura, Takayuki Yoshino

Disclosures

Yoshiaki Nakamura: Chugai Pharmaceutical Co. Ltd. (H); Takayuki Yoshino: Taiho Pharmaceutical Co., Ltd., Chugai Pharmaceutical Co., Ltd., Eli Lilly Japan K.K. (H), GlaxoSmithKline plc., Nippon Boehringer Ingelheim Co., Ltd. (RF).

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board