Clinical applications of cell-free DNA-based liquid biopsy analysis

Highlights

• •ctDNA testing enables identification of actionable mutations for targeted therapy.
• •NGS-based ctDNA testing has been applied in genomic profiling of cancer patients.
• •ctDNA testing has been applied to cancer treatment monitoring and MRD detection.
• •Clinical application of ctDNA assays requires understanding of their pros and cons.
• •Minimal residual disease-based treatments can impact future cancer treatments.

Abstract

Liquid biopsies, particularly those involving circulating tumor DNA (ctDNA) from patient blood, have emerged as crucial and minimally invasive adjuncts to standard tissue-based testing. ctDNA testing enables the identification of actionable mutations for targeted therapy and can be routinely used when tissue samples are unavailable for genotyping. Compared to tissue-based testing, ctDNA testing has the advantages of capturing spatial or temporal genomic heterogeneity and facilitating repeated assessments. The utility of liquid biopsies extends to multiple clinical applications, including cancer diagnosis, treatment monitoring, and minimal residual disease (MRD) detection. Numerous clinical trials are currently evaluating treatment strategies using ctDNA testing. In particular, the implementation of adjuvant treatment escalation or de-escalation based on MRD detection could dramatically transform future approaches to solid tumor treatment. Various ctDNA assays have been developed, and it is important to understand their strengths and weaknesses for effective clinical applications. Furthermore, ctDNA testing faces several technical challenges, including low sensitivity in detecting copy number alterations and fusions, as well as the possibility of detecting mutations associated with clonal hematopoiesis of indeterminate potential. In this review, we comprehensively discuss the methodologies and recent advancements in cfDNA-based liquid biopsies for cancer patients, covering diagnosis, genomic profiling, and treatment monitoring. Furthermore, we explore clinical trial designs employing ctDNA testing and anticipate forthcoming changes in patient care.

Introduction

Molecular profiling of tumors is essential for optimizing treatment. In recent years, next-generation sequencing (NGS) techniques have been used to determine treatments based on the exact genomic profile of a particular tumor. Current clinical practice often involves sequencing previous primary tumor tissues obtained using surgery or biopsy. However, tissue biopsy requires an invasive procedure, may not provide sufficient samples, and cannot be performed in multiple metastatic diseases and all organs. Moreover, genetic heterogeneity is not considered when examining single-site tissue biopsies, which is an important limitation because genetic profiles often differ between primary tumors and metastases [1]. Tumor cells also acquire genetic alterations related to resistance during treatment because of clonal evolution [2]. To overcome these problems, liquid biopsy, particularly circulating tumor DNA (ctDNA) analysis, has been actively investigated. In the 1990s, tumor-derived genomic alterations were found to be detectable in "cell-free DNA" (cfDNA) circulating in the plasma of patients with malignant tumors [3]. Circulating nucleic acids can be detected in the blood, urine, and cerebrospinal fluid, and their diagnostic assessment in liquid specimens is commonly referred to as liquid biopsy [4]. ctDNA has emerged through the detection of small quantities of tumor DNA in peripheral blood. Additionally, the cfDNA-based liquid biopsy assay has been used to identify genetic alterations for targeted therapy in clinical practice [5] as well as for potential clinical applications in cancer screening, diagnosis, minimal residual disease (MRD) detection, and real-time monitoring of tumor evolution and resistance (Fig. 1). Identification of ctDNA in cfDNA is a technical challenge, and researchers are actively developing detection methods [6]. This approach is expected to lead to diverse clinical applications. Treatment strategies using liquid biopsy technology are still recent, and many challenges need to be resolved. Here, we review the status and potential clinical applications of cfDNA-based liquid biopsy assays. While we acknowledge the value of liquid biopsies using other tumor-derived products, such as circulating tumor cells, circulating cell-free RNA, or extracellular vesicles, this particular review will not cover them.

Fig. 1.

Clinical applications of genome-wide cfDNA fragmentation in cancer patients. cfDNA can be analyzed to identify several DNA-based alterations in cancers, including mutations, copy number alterations, translocations, and DNA methylation changes.

Biology of cfDNA

cfDNA is released from normal and cancer cells during necrosis and apoptosis [7]. During cell death, nucleases that cleave nucleosome fragments cause DNA fragmentation. This fragmentation is not random, and mapping of the ends of cfDNA fragments can reflect the nucleosome profile of the cells from which it is derived. Therefore, this fragmentation can be used to distinguish cfDNA from normal and cancerous cells. In addition, genome-wide analysis of nucleosome-occupied regions—called nucleosome footprints—has revealed that it is possible to infer the tissue of origin of cfDNA, and a characteristic pattern has been observed in cancer patients [8]. DNA fragments released from normal tissues or tumors exhibit varying lengths. The peaks representing the size of DNA fragments derived from normal tissues were found to be approximately 166 bp, while those originating from ctDNA were approximately 143–145 bp [8,9]. Additionally, the concentration of cfDNA in healthy subjects was 1–10 ng/mL, while it was higher in patients with advanced non-small cell lung cancer (NSCLC), at approximately 100–180 ng/mL [10]. cfDNA from normal cells in healthy individuals corresponds to the length of DNA wrapped around nucleosomes (approximately 180 bp). Only a small fraction of cfDNA is considered ctDNA (<1%) [11], and the clearance time of DNA derived from normal tissue or ctDNA is different [12].

For the detection of ctDNA, the pre-analytical conditions of patient-derived specimens may affect the assay results. The quality of the collection tube, storage conditions, sample volume, and effects of white blood cell lysis should be carefully monitored. In general, at least 20 mL of whole blood should be collected to ensure sufficient plasma DNA for analysis, but this depends on the nature of the assay. Ideally, cfDNA should be isolated within a few hours of blood collection to prevent leukocyte lysis. Otherwise, a large amount of cellular genomic DNA will be released into the plasma components in the collection tube due to cell lysis. Instead of standard EDTA tubes, a more specific cfDNA blood collection tube may be an option to prevent cellular lysis and stabilize the cell membrane [13]. Due to the low variant allele frequency, for accurate detection of ctDNA, more sensitive and specific types of assays should be utilized, depending on the purpose [14]. Detection of single nucleotide variants requires more sensitive assays, and loss of heterozygosity and copy number gains or losses are technically difficult [14].

ctDNA assay

Methods for the detection of ctDNA are classified into polymerase chain reaction (PCR)-based and NGS-based techniques. Digital PCR, the initial method of single-molecular amplification, can identify selected rare sets of genomic alterations at a low cost [15]. BEAMing (beads, emulsion, amplification, and magnetics) [16] and droplet digital PCR [17] (ddPCR, including multiplex ddPCR) are the most prevalent digital PCR-based methods for the assessment of ctDNA. BEAMing enables the conversion of single DNA molecules into single magnetic beads and quantification by counting stained beads containing amplified DNA molecules using flow cytometry. A good agreement between BEAMing and ddPCR has been reported, suggesting sufficient reproducibility for clinical use [18]. Digital PCR fulfills the investigation requirements with a shorter turnaround time and allows a limit of detection of 0.001–0.01 % [19]. However, the number of fluorescent probes used in one assay can be up to five; thus, it cannot acquire a wide range of genomic information. These single-locus assays are best suited for detecting “hotspot” mutations and small indels with high sensitivity.

NGS-based techniques utilize amplicon-based PCR or hybrid capture-based methods to cover hundreds of genes. In the past, using NGS for ctDNA analysis was hindered by the fact that NGS is prone to sequencing errors. Several breakthrough techniques, such as hybridization capture from sequencing libraries using target-specific probes or direct PCR amplification of the target-specific primer, allow for ultrasensitive detection of ctDNA [[20], [21], [22]]. PCR amplicon methods can cover around 100 genes with high sensitivity. In contrast, hybrid capture methods can cover a wider range of genomic regions than amplicon-based methods and detect not only point mutations but also small indels, structural variants and copy number alterations.

More comprehensive approaches, including whole-exome sequencing (WES) and whole-genome sequencing (WGS), can capture broader genomic changes. Low-coverage WGS has been used to detect cancer-specific genome-wide copy number alterations and mutational signatures [23]. Despite the low detection sensitivity of genomic abnormalities, low-coverage WGS is in the process of overcoming its main limitations [24]. Table 1 summarizes ctDNA analysis methods and their characteristics.

Table 1.

Methods of ctDNA analysis.

Target	Methods	Technique	Characteristics	Product (example)	Limit of detection
Single locus	Quantitative PCR	ARMS-PCR	Detection of specific hotspot mutation	EGFR mutation test (cobas®, Biocept®) therascreen PIK3CA RGO PCR kit (Qiagen®)	0.01–0.1 %
Single locus	Digital PCR	ddPCR BEAMing	Detection of specific hotspot mutation with quantification	OncoBEAM RAS CRC(Sysmex®)	0.01 %
Target gene	PCR amplicon-based NGS	Safe-Seq TAm-Seq	Cover range is narrow, low DNA input	Signatera	0.02–0.1 %
	Hybrid capture-based NGS	CAPP-Seq	Cover range is wide, high DNA input Detection CNV and fusion	FoundationOne Liquid Guardant 360	0.004–0.02 %
Comprehensive	whole exome or genome sequence		Methylation, fragmentation analysis Mutational signature, genome-wide CNAs		1–10 %

*Abbreviations: beads, emulsion, amplification, and magnetics (BEAMing), droplet digital PCR (ddPCR), next-generation sequencing (NGS), polymerase chain reaction (PCR), Safe-Sequencing System (Safe-Seq), tagged-amplicon deep sequencing system (TAm-Seq).

Genomic profiling in metastatic cancer

ctDNA is useful for detecting genetic aberrations in patients with metastatic solid tumors who cannot undergo biopsy, and it can also be used as a guide for treatment selection in metastatic cancer. Initially, PCR-based assays were developed as companion diagnostic tests to detect druggable targets. PCR-based assays for oncogenic driver mutations, such as EGFR, have high sensitivity and specificity [25].

In recent years, NGS-based multigene assays have become more commonly used than PCR-based single-gene assays. Recent prospective studies of breast, lung, and gastrointestinal cancers have shown that NGS-based ctDNA genotyping accurately identifies mutations and can be used to guide treatment decisions [[26], [27], [28]]. Today, ctDNA testing can be used to detect all guideline-recommended and treatable oncogenic drivers in treatment-naïve metastatic NSCLC when tissue specimens are unavailable [29]. ctDNA-based genotyping not only provides a useful alternative to tissue-based genotyping but also offers faster results, thereby improving access to targeted therapy. For example, the SCRUM-Japan GI-SCREEN and GOZILA studies assessed the differences in clinical trial enrollment between tissue-based and ctDNA-based genotyping in over 7000 patients with gastrointestinal cancer. ctDNA genotyping significantly shortened the turnaround time (11 vs. 33 days, P < 0.001) and improved the matched trial enrollment rate (9.5 vs. 4.1 %, P < 0.001) without compromising the treatment efficacy compared to tissue genotyping [28]. ctDNA can also provide biomarkers for immunotherapy, including tumor mutational burden (TMB) and microsatellite instability (MSI). TMB is the number of somatic mutations in the tumor genome, which reflects the neoantigen load in predicting the response to immune checkpoint inhibitors (ICIs). The FDA approved the FoundationOne CDx assay as a companion diagnostic tool for pembrolizumab in TMB-high (≥10 mut/Mb) solid tumors. The standard approach for the analysis of TMB is based on WES of tumor biopsy, and a high concordance has been reported between WES and tissue-based panel tests [30]. Several studies have suggested that blood-based TMB (bTMB) could serve as a surrogate for tissue-based TMB (tTMB) [[31], [32], [33]]. These studies reported a moderate correlation between bTMB and tTMB (Spearman rank correlation = 0.64–0.74) when adequate levels of plasma tumor fraction were present, and bTMB could predict the response to ICIs [32,33]. However, it is important to interpret bTMB results with caution owing to potential influences from differences in tumor fractions, computational algorithms used in the analysis, intratumoral heterogeneity, and sample processing quality. For example, results from a study evaluating concordance between bTMB and tTMB in 410 patients with solid tumors showed that the median bTMB was significantly higher than that of tTMB (10.5 vs. 6.0 mut/Mb), leading to conflicting TMB-high status results in over one-third of cases [34]. At this time, no prospective trials have demonstrated the utility of bTMB as a predictive marker of ICI response. Recent findings from the BFAST trial, which evaluated atezolizumab or chemotherapy in patients with first-line metastatic NSCLC and bTMB-high (14.5 mut/Mb), showed no difference in progression-free survival (PFS) between the treatment arms [35].

Several trials have revealed that MSI detection using ctDNA assays is highly concordant with tissue-based testing. For example, in a study examining over 1000 patients with various cancers, ctDNA MSI evaluation accurately detected 87% of tissue MSI-high, with a detection limit of 0.1% tumor content. In clinical trials of pembrolizumab in patients with advanced gastric cancer, 10 of the 16 patients with MSI-high detected by ctDNA achieved an objective response [36]. Furthermore, MSI detection using ctDNA was better for samples with higher ctDNA fractions (≥1%) [37].

Early assessment of treatment efficacy

ctDNA is useful for real-time monitoring of efficacy during systemic treatment. Some studies have shown that the amount of ctDNA is related to the tumor burden [22,38]. ctDNA increases several months before tumor marker elevation or clinical disease progression and declines faster than any other marker once treatment is effective [22]. Moreover, changes in ctDNA levels during treatment are associated with prognosis across various cancer types and treatment contexts [[39], [40], [41], [42], [43]]. One area of focus is ICI treatment. Relying solely on initial clinical and radiologic responses can be misleading due to delayed responses or pseudo-progression. Several studies have revealed that measuring early changes in ctDNA levels after ICI treatment (6–9 weeks) was predictive of ICI benefits, such as RECIST response and overall survival [44,45]. Patients with decreased variant allele frequency after treatment were associated with longer survival than those with increased frequency. Moreover, patients who cleared their ctDNA levels during treatment had the most favorable survival outcomes. For example, among 73 patients with advanced solid tumors treated with pembrolizumab, 12 patients who achieved ctDNA clearance were still alive at the end of the study, with a median of 25 months follow-up. Additionally, ctDNA clearance often occurred before the radiologic response. One patient in the same study achieved ctDNA clearance by four months, while a radiologic response was not observed until eight months after treatment [45]. Another study evaluating serial ctDNA profiles in patients with metastatic melanoma treated with ICIs showed that changes in ctDNA during treatment can accurately differentiate pseudo-progression from true disease progression [46]. Further studies are needed to define the optimal timing of measurement and the threshold for quantitative changes.

Understanding resistance mechanisms

Understanding acquired genetic abnormalities after treatment of patients with metastatic cancer may help evaluate treatment efficacy. ctDNAs are suitable for monitoring clonal evolution and detecting resistance mutations after drug exposure. Commonly acquired resistance mutations include RAS/BRAF/EGFR mutations from anti-EGFR antibodies, EGFR T790M mutations from EGFR-TKIs, ESR1 mutations from aromatase inhibitors (AIs), and BRCA1/2 reversion mutations from platinum chemotherapy and PARP inhibitors [47,48]. Treatment intervention through the detection of resistance mutations using ctDNA has shown clinical benefits. For example, in the CHRONOS trial, patients with metastatic colorectal cancer who had tissue-RAS wild-type status after previous treatment with anti-EGFR-based regimens underwent ctDNA-based screening to guide anti-EGFR rechallenge decision [49]. Patients with no detectable alterations in RAS, BRAF, and EGFR extracellular domain mutations in ctDNA were subsequently re-treated with an anti-EGFR antibody, panitumumab. Of the 27 enrolled patients, eight (30%) achieved a partial response. Another example is the PADA-1 trial, which evaluated whether treatment modification by ESR1 mutations would benefit patients with advanced breast cancer treated with AIs and CDK4/6 inhibitors [50]. Patients with increased ESR1 mutations and without clinical disease progression were randomized to continue with AIs or switch to fulvestrant and a CDK4/6 inhibitor. Patients who switched to fulvestrant exhibited improved PFS compared with those who continued with AIs. Further studies are needed to develop optimal panels for each cancer type and the optimal monitoring frequency.

MRD detection

ctDNA is used for detecting MRD and identifying patients at a high risk of recurrence. MRD is characterized by a small number of cancer cells remaining after curative treatment that cannot be detected using clinical imaging modalities, leading to recurrence. As the blood half-life of ctDNA is several hours, ctDNA detected after curative treatment may indicate the presence of MRD. In studies where ctDNA was obtained in parallel with standard surveillance, there was a lead time of 3–18 months between ctDNA-based MRD detection and radiologic recurrence [[51], [52], [53], [54], [55], [56], [57], [58]]. Adjuvant therapy may be modified based on the presence or absence of MRD. For example, patients with MRD should receive adjuvant therapy, and patients without MRD can be omitted from therapy to avoid unnecessary adverse events [59].

To accurately detect MRD in solid tumors, it is necessary to identify genetic characteristics specific to the target tumor and establish a sensitive detection method. Currently, several MRD panels have been developed and classified into two main types: tumor-informed and tumor-naïve. Tumor-informed panels can identify mutations in tumor tissues and track patient-specific mutations for high sensitivity. Deep sequencing has been achieved by reducing background noise from non-tumor-derived mutations [60]. Tumor-naïve panels are considered less sensitive because they detect de novo mutations from plasma; however, they may detect emerging resistance mutations [61]. The specificity of ctDNA detection for predicting recurrence was high across ctDNA assays and cancer types. For example, in patients with stage II/III breast cancer enrolled in the I-SPY2 trial. ctDNA was detected by a tumor-informed panel in 6 of 60 (10 %) patients after neoadjuvant chemotherapy (NAC), 5 of whom (83.3%) showed recurrence. In contrast, recurrence occurred only in 6 (11.1%) of the 54 patients with negative ctDNA results (HR 11.28, P < 0.0001). Although pathological complete response (pCR) to NAC is strongly associated with a better prognosis in breast cancer, this study also found that the risk of distant recurrence in patients with ctDNA-negative despite not achieving pathological complete response (non-pCR) was similar to that in those who achieved pCR [62]. Parikh et al. investigated the use of a tumor-naïve panel [63] to detect MRD in colorectal cancer patients using NGS with somatic and cancer-specific methylation signatures. Of the 70 patients studied, 17 (24%) were found to be ctDNA-positive after completion of definitive therapy. Notably, all 15 (100%) ctDNA-positive patients experienced recurrence within at least one year of follow-up. Of the 49 patients with negative ctDNA, 12 (24%) showed recurrence (HR 11.28, P < 0.0001). Integration of ctDNA methylation with genomic alteration analysis improved panel sensitivity from 40.7% to 55.6%. To date, an association between MRD and prognosis has been reported for various cancer types [[64], [65], [66], [67], [68], [69], [70], [71]]. We have summarized the results of observational studies evaluating the association between MRD detection using ctDNA and prognosis in various types of cancer in Fig. 2. These results showed that patients with MRD are more likely to relapse and have poor prognoses, while it remains largely unclear whether MRD detection using ctDNA can guide clinical decision making. Although prospective data are limited, recent results from a DYNAMIC trial demonstrated that a ctDNA-guided approach reduces the use of adjuvant chemotherapy in stage II colon cancer without compromising the recurrence risk. Patients in the ctDNA-guided group received a lower percentage of adjuvant chemotherapy than those in the standard group (15% vs. 28%, P = 0.0017), while achieving non-inferior 2-year recurrence-free survival (93.5% vs. 92.4%, absolute difference: +1.1 %; 95 % CI −4.1–6.2) [72]. These data were extremely important in the development of individualized approaches to postoperative treatment and are expected to be applied to other types of cancer.

Fig. 2.

Forest plot of the hazard ratios for recurrence-free survival or disease-free survival in patients who were positive for ctDNA.

Cancer diagnosis

ctDNAs have been investigated as useful tools for cancer diagnosis. Although early detection strategies based on ctDNAs are promising, numerous obstacles must be overcome before they can be applied clinically. The detection rate of ctDNA varies depending on the tumor volume [22,38] and the anatomical site of disease [11,73]. For example, in a study involving 640 patients with various types of cancer, the ctDNA detection rate was examined and found to be associated with the stage of cancer: stage I (47%), II (55%), III (69%), and IV (82%) [11]. Additionally, the detection rate of ctDNAs varies depending on the organ. Fewer than10 % of patients with brain tumors harbor detectable ctDNA. Moreover, clonal hematopoiesis of indeterminate potential (CHIP) mutations affect test specificity [74]. Although some methods, including integrated genetic and protein biomarkers, fragmentation profiles, and methylation-based ctDNA assays, have been investigated for early cancer detection, their sensitivity might be a limitation for clinical applications. For example, the median sensitivity of CancerSEEK, an integrated approach that combines genetic alterations and protein biomarkers, for detecting non-metastatic cancers was reported as 70% (ranging from 33% to 98%) [75]. As another example, targeted methylation analysis of circulating cfDNA showed that the sensitivity and specificity for discriminating cancer from non-cancer were 51.5% and 99.5%, respectively [76]. The accuracy of primary origin detection in these assays was over 80% [75,76]. Well-designed prospective studies are needed to assess cfDNA as a screening tool for patients in whom cancer screening is recommended.

Caveats and technical challenges of ctDNA analysis

A liquid biopsy is considered less invasive, easily repeatable, and more likely to represent the whole picture of genomic heterogeneity. However, there are some caveats to ctDNA analysis.

Less sensitivity for fusion detection

Although gene fusions are considered an important marker for diagnosis or treatment, their detection is technically challenging in liquid biopsies. As a conventional single method, fusion detection in situ hybridization is one of the most widely available techniques. It is difficult to identify the exact variant fusion pairs using this assay. For detection using a targeted NGS assay, RNA-seq is considered a more sensitive method because various fusion types result in the same oncologic transcript [77]. Specifically, DNA analysis is interfered with by large introns or repetitive sequences, and it is difficult to discriminate the true expression of fusion genes. The detection of gene fusions using plasma NGS is more variable than that observed in point mutations and indels [78]. Although large-scale studies remain limited, a study of 137 patients with a fusion detected in tissue showed that the sensitivities of ctDNA for ALK and RET fusion detection were 73% and 71%, respectively. The sensitivity of fusion detection is associated with the ctDNA fraction, and the sensitivities were 85% and 52% for ctDNA specimens with ctDNA fractions ≥1% and <1%, respectively [79]. Given the modest sensitivity of ctDNA for gene fusion detection, reflex tissue testing should be considered when available.

Clonal hematopoiesis

CHIP, a process that involves the accumulation of somatic mutations in hematopoietic stem cells, leads to the clonal expansion of mutations in blood cells [80]. CHIP is part of the normal aging process and is highly prevalent in the general population. An adult human has been estimated to have approximately 50,000–200,000 stem cells [81], and an average person would potentially harbor up to 1.2 million exonic mutations by the age of 70 [80]. According to Razavi et al., up to 10 % of clonal hematopoiesis mutations detected in the plasma were listed as oncogenic in the OncoKB database, and 13 % of these mutations were indicated for an approved targeted therapy or clinical trial [82]. Therefore, caution is required because incorrect identification of alterations leads to inappropriate treatment selection. Although sequencing of white blood cell DNA could exclude or minimize the possibility of CHIP, most available assays only analyze plasma. Thus, detecting oncogenic-driver variants in plasma-only ctDNA assays requires careful interpretation, especially when variants potentially associated with CHIP (e.g., TP53, ATM, and KRAS) are detected in patients over 70 years old.

Future directions for clinical application

To date, ctDNA testing has proven useful not only for genomic profiling but also for predicting prognosis and treatment efficacy. However, most evidence on MRD detection and response monitoring is based on retrospective studies, and it remains unclear whether ctDNA-based treatment strategies are clinically beneficial. Several intervention studies using ctDNA testing are underway, and examples of clinical trial designs have been discussed [83]. The most straightforward study design is to add systemic therapy to patients who are MRD-positive after completing standard curative treatment (Fig. 3A), which may eliminate micrometastasis and prevent recurrence. Serial ctDNA testing predicts recurrence with a sensitivity of 79–100% and specificity of 89–100%, allowing more efficient enrollment of patients at high risk of recurrence [84]. The lead time from MRD detection to radiologic recurrence has been reported to be 8.9–12.4 months for early-stage breast cancer [51,54,85], depending on the ctDNA assay, the interval between imaging tests, and disease biology. Thus, it is crucial to select appropriate patient populations, assays, and ctDNA testing intervals while also performing sensitive imaging studies before study enrollment. For example, the c-TRAK TN study focused on triple-negative breast cancer patients with residual disease after neoadjuvant chemotherapy, who were at extremely high risk of recurrence, and assessed the utility of prospective ctDNA surveillance. Of the patients with MRD, 70% already had metastases on imaging simultaneously [86]. Furthermore, early identification of recurrence by combining ctDNA testing with surveillance of radiologic imaging after curative treatment may influence patient prognosis and treatment strategies (Fig. 3D). For locoregional recurrence or oligometastatic disease, adding local treatment with surgery or radiation therapy may improve prognosis [87,88].

Fig. 3.

Ongoing clinical trials designed using ctDNA. (A) Escalation strategies for early-stage solid tumors. Patients who were positive for ctDNA after standard perioperative treatment were randomized to receive additional experimental treatment or observation. (B) Treatment monitoring for advanced solid tumors. Patients with arising resistance mutations or elevated ctDNA levels before clinical progression were randomly assigned to the switch/add-on treatment group and the continue treatment group. (C) De-escalation strategies for early-stage solid tumors. Participants were randomly assigned to treatment ctDNA-guided (ctDNA-positive patients received standard adjuvant therapy, and negative cases were observed) versus control (standard chemotherapy or observation based on standard clinical criteria) groups. (D) Surveillance for recurrence. Patients were randomly assigned to the ctDNA-based surveillance (CT scan when ctDNA became positive) versus standard surveillance according to the guidelines.

Next, we discuss the study designs for MRD-negative patients. In breast and colorectal cancer, researchers have explored risk stratification based on clinical or genetic factors to omit or shorten the duration of adjuvant chemotherapy. For instance, in patients with hormone receptor-positive breast cancer deemed to be at low to intermediate risk for recurrence according to gene expression assays, such as Oncotype DX and ManmaPrint, adjuvant chemotherapy can be safely omitted without worsening the prognosis [[89], [90], [91]]. Moreover, MRD detection by ctDNA testing may help omit unnecessary systemic therapy. Clinical trials can be conducted using two main study designs: (1) randomizing MRD-negative patients to standard treatment versus observation, and (2) assigning postoperative treatment based on MRD testing (standard adjuvant therapy for MRD-positive patients and no treatment for MRD-negative patients) or clinicopathologic factors (Fig. 3C). In the latter scenario, the feasibility of omitting standard adjuvant treatment for MRD-negative patients cannot be definitively determined. Furthermore, MRD negativity does not preclude recurrence. In fact, in the DYNAMIC II trial, 6% of MRD-negative patients without adjuvant chemotherapy developed recurrence [72]. Therefore, careful consideration should be taken to ensure that the false-negative rate of the assay is sufficiently low. Notably, intracranial and intra-abdominal recurrent tumors exhibit low shedding into the plasma, which may lead to false-negative results [92,93]. Confirmation of MRD negativity by serial ctDNA testing, in combination with other clinical factors, may decrease the false-negative rate. In addition, MRD detection by ctDNA testing may answer the question of the appropriate duration of postoperative treatment. For example, although postoperative endocrine therapy for early breast cancer is recommended to be continued for more than five years, there is no clear guidance on the duration of treatment. It may be possible to complete endocrine treatment if MRD is negative ≥5 years after initiation of treatment. Ongoing studies evaluating the utility of MRD detection are outlined in Table 2.

Table 2.

Ongoing interventional studies evaluating the utility of MRD detection using ctDNA in solid cancers.

Trial	Arms and Interventions	Study	Clinical trial number	Cancer Type	Stage / Status	Detection Methods	Platform	Outcome
ctDNA-Guided De-escalation Strategy Design	Randomly assigns participants to treatment ctDNA guided (ctDNA positive receive standard adjuvant therapy, negative cases are observed) versus control (standard chemotherapy or observation based on standard clinical criteria)	DYNAMIC II DYNAMIC III DYNAMIC RECTAL TRACC MEDOCC—CrEATE COBRA	ACTRN12615000381583 ACTRN12617001566325 ACTRN12617001560381 NCT04050345 NL6281/NTR6455 NCT0406810	Colon Colon Rectal Colorectal Colon Colon	Stage II Stage III Locally advanced High risk stage II, stage III Stage II Stage IIA	Safe-SeqS Safe-SeqS Safe-SeqS Amplicon NGS Capture NGS Capture NGS	NA NA NA Signatera™ PGDx elio™ Guardant LUNAR-1™	DFS in the ctDNA-positive patients
	Patients who are negative for postoperative ctDNA are randomized to adjuvant chemotherapy or to follow-up.	VEGA	jRCT1031200006	Colon	High-risk stage II, low-risk stage III	Amplicon NGS	Signatera™	To demonstrate the non-inferiority of observation in ctDNA negative patients
ctDNA-Guided Escalation Strategy Design	Patients who are positive for postoperative ctDNA are randomized to adjuvant chemotherapy/experimental therapy or to follow-up/standard therapy.	CIRCULATE AIO-KRK-0217 CIRCULATE PRODIGE 70 LEADER ZEST IMvigor011	NCT04089631 NCT04120701 NCT03285412 NCT04915755 NCT04660344	Colon Breast Breast Bladder	Stage II Stage II HR+/HER2-: T1c-T4c, any N, Stage I-III TN: tBRCAwt, HR+/HER2-: tBRCAmut Muscle-invasive	Not mentioned ddPCR (2 methylated markers WIF1 and NPY) Amplicon NGS Amplicon NGS Amplicon NGS	NA NA Signatera™ Signatera™ Signatera™	To demonstrate the superiority of adjuvant chemotherapy/experimental therapy in ctDNA-positive patients
	Patients who are ctDNA-positive after completion of adjuvant chemotherapy are randomized to: standard of care (placebo/surveillance) or to experimental treatment.	ALTAIR ACT-3 DARE TRAK-ER c-TRAK-TN	JapicCTI-205,363/NCT04457297 NCT03803553 NCT04567420 NCT04985266 NCT03145961	Colorectal Colorectal Breast Breast Breast	Stage II-III or stage IV with resectable metastases Stage III Stage II-III HR+/HER2- High risk HR+/HER3- Moderate or high risk early-stage triple negative	Amplicon NGS Capture NGS Amplicon NGS NA Droplet digital PCR	Signatera™ Guardant LUNAR-1™ Signatera™ Not mentioned BioRad QX200™	To demonstrate the superiority of experimental treatment over standard of care in patients with positive ctDNA after standard adjuvant therapy
ctDNA-Guided Surveillance Strategy Design	Patients who are ctDNA-positive during surveillance are followed-up by FDG-PET/CT.	IMPROVE-IT2	NCT03748680	Colorectal cancer	Stage I-II	Droplet digital PCR (colorectal panel)	NA	To demonstrate that ctDNA guided postoperative surveillance combining ctDNA and radiological assessments could result in earlier detection of recurrent disease and identify more patients eligible for curative treatment

Finally, the study design could include treatment switching if treatment monitoring reveals resistant mutations or increasing ctDNA levels (Fig. 3B). Patient outcomes may be improved by switching treatment before evidence of radiological progression. However, further study, including cost-effectiveness analyses, is needed to determine the population that would benefit most from these treatment strategies utilizing ctDNA monitoring. For example, the PADA-1 trial [50] showed a prolonged PFS with treatment switching based on ESR1 mutation detection, whereas the PFS improvement, including crossover treatment of the control arm, was only 2.7 months. Further, ESR1 mutations without radiographic disease progression were identified in only 19% of patients. Thus, <20% of patients who underwent ctDNA monitoring benefited from the treatment strategy. ctDNA dynamics are an early predictor of response to chemotherapy and immunotherapy and can be used to optimize treatment. In breast cancer, ctDNA changes during neoadjuvant treatment are a predictor of pCR [62,94]. Hence, early identification of non-responders and treatment modification may improve the pCR rate and prognosis [95]. However, it is not yet clear whether ctDNA dynamics and clearance are surrogate endpoints for long-term outcomes, and further validation through clinical studies is warranted.

While large-scale genomic data are being accumulated, mainly in Western countries, genomic data in Asia and other regions are still limited. The incidence and histology of some cancer types vary according to race and region. In Asia, the incidence of nasopharyngeal cancer, cervical cancer, and ovarian clear cell carcinoma is higher than in Western countries. Among patients with breast cancer, the frequencies of TP53 and PIK3CA mutations, which are major driver mutations in breast cancer, differ according to race [96]. The distribution of molecular subtypes also varies according to race [97]. These results suggest that treatment response and prognosis may differ based on race, and it is important to explore race-specific biological characteristics. We are currently conducting the A-TRAIN study, ctDNA-based observational study on Asian patients (NCT05099978), which includes cohorts of cervical cancer, ovarian clear cell carcinoma, and nasopharyngeal cancer. In the future, this study will clarify the carcinogenesis and genomic profiling of common cancers in Asia and will explore race-specific treatment strategies in Asian patients with cancer.

Conclusions

Liquid biopsy has already demonstrated its value in genotyping advanced cancer and holds potential for a wider range of clinical applications, including cancer diagnosis, treatment effect monitoring, and MRD detection. The ongoing clinical trials investigating its clinical utility are poised to influence a transformative shift in patient care. To maximize its effectiveness, the development of an optimal panel is crucial, since target gene abnormalities and required detection sensitivity differ depending on the cancer type and purpose of detection. Although the cost of this method remains a challenge, further technological innovations are warranted for addressing this concern and realizing the full potential of liquid biopsy in precision oncology.

Funding information

This work was supported by a Grant-in-Aid for Scientific Research (B) (grant number 20H03695), a Japan Agency for Medical Research and Development (23ama221520h0001), and the National Cancer Center Research and Development Fund (grant number 2020-J-2 and NCC Core Facility).

CRediT authorship contribution statement

Shu Yazaki: Conceptualization, Methodology, Investigation, Data curation, Writing – original draft, Visualization, Project administration. Momoko Tokura: Conceptualization, Methodology, Investigation, Data curation, Writing – original draft, Visualization. Hisaki Aiba: Conceptualization, Methodology, Investigation, Data curation, Writing – original draft, Visualization. Yuki Kojima: Conceptualization, Methodology, Investigation, Data curation, Writing – original draft, Visualization. Kouya Shiraishi: Conceptualization, Methodology, Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.