Clinical application of molecular residual disease detection by circulation tumor DNA in solid cancers and a comparison of technologies: review article

ABSTRACT

Molecular residual disease (MRD), detected by circulating tumor DNA (ctDNA) can be involved in the entire process of solid tumor management, including recurrence prediction, efficacy evaluation, and risk stratification. Currently, the detection technologies are divided into two main categories, as follows: tumor-agnostic and tumor informed. Tumor-informed assay obtains mutation information by sequencing tumor tissue samples before blood MRD monitoring, followed by formulation of a personalized MRD panel. Tumor-agnostic assays are carried out using a fixed panel without the mutation information from primary tumor tissue. The choice of testing strategy may depend on the level of evidence from ongoing randomized clinical trials, investigator preference, cost-effectiveness, patient economics, and availability of tumor tissue. The review describes the difference between tumor informed and tumor agnostic detection. In addition, the clinical application of ctDNA MRD in solid tumors was introduced, with emphasis on lung cancer, colorectal cancer, Urinary system cancer, and breast cancer.

1. Introduction

Globally, cancer is one of the leading causes of death. According to the Global Cancer Observatory 2020 global cancer statistics, there are approximately 10 million cancer deaths each year.¹ Especially those patients relapsed after surgical resection or drug therapy would suffer from the decreased survival and quality of life. Therefore, better biomarkers are needed to identify patients at risk of recurrence. Many investigators believe that these relapse cases have a small number of uncleared tumor cells, even if the lesions have been cleared on imaging after treatment.^2,3 Molecular Residual Disease (MRD), which can also be considered as minimal residual disease or measurable residual disease, is discussed in this review for residual tumors at the molecular level that cannot be detected by conventional imaging methods.⁴ To date, MRD testing has made excellent progress as a tumor biomarker that can provide more detailed individualized decision during cancer management, including early detection of recurrence and prediction of treatment efficacy and prognosis (Figure 1).

Figure 1.

The schematic of key time points for ctDNA MRD detection (a) and the significance of MRD status in predicting recurrence, prognosis and guide clinical decision-making (b).

The baseline for MRD detection was the time point prior to any antitumor therapy. The landmark times were 7 days to 1 month after the definitive treatment completed. The longitudinal points of follow-up commonly refer to every three to 6 months since the landmark point.

Currently, circulating tumor DNA (ctDNA) and circulating tumor cells assays in the blood are used to monitor the MRD in peripheral blood and other body fluids.⁵ For example, cerebrospinal fluid is used for brain tumors, saliva is used for laryngeal cancers, hydrothorax, and ascites are used for metastatic cancers of the chest and abdomen, and urine is used for cancers of the urinary tract.⁶ Compared with tissue biopsy, liquid biopsies offer the advantages of being minimally invasive, objective, efficient, and continuously follow a patient’s status during treatment.^5,7–9 However, ctDNA accounts for less than 1% of circulating free DNA (cfDNA),¹⁰ which existing in body fluids and is approximately 200 base pairs.¹¹ Therefore, with the establishment of PCR- and NGS-based molecular analysis methods, the sensitivity and lower limit of detection (LOD) for tumor-derived DNA in blood have improved.¹²

This review summarizes the main detection technologies, tumor informed and tumor agnostic methods of MRD, and provides a review of clinical applications of MRD for lung cancer, colorectal cancer, breast cancer, and genitourinary cancer. In addition, the challenges and prospects of MRD applications are presented.

2. MRD analysis strategy

Despite the low levels of ctDNA in blood, the sensitivity and lower limits of detection of tumor-derived material in blood have changed with the establishment of PCR and NGS-based molecular analysis methods. Currently, MRD detection technologies are divided into tumor agnostic and tumor informed methods. These two techniques were often performed in recent studies, but there is a lot of debate about the two techniques. We would like to describe the technical methods and performance of these two representative methods as follows.

2.1. Tumor-informed assays

Generally, for tumor-informed assay, tumor genomic landscape was firstly analyzed by sequencing tumor biopsies or surgical samples [e.g., whole exome sequencing (WES) or large gene panels, or tumor-specific panels], and then a personalized gene panel was designed for blood monitoring by ultra-depth of PCR/NGS.¹³ Besides, for plasma sequencing, there is no standard in the sequencing depth, detection sensitivity or threshold for ctDNA positive.

The current FDA approved commercial ctDNA MRD products are designed based on tumor-informed analysis strategies. As an example, Signatera (Natera Inc., Austin, TX, USA) performs WES of tumor tissue to identify somatic variants, then a personalized panel including 16 somatic variants is used for ctDNA detection and tracking.^14,15 The average sequencing depth for ctDNA-specific variant detection is up to 100,000 ×, with a detection limit of 0.001% VAF^2,16 and a cutoff for ctDNA positivity is defined as the presence of ≥ 2 somatic variants.¹⁷ In a study, 90 tumor specimens and 363 serial plasma samples from NSCLC patients were collected and tissue sequencing of WES was also performed, and a patient-specific panel including 48 variants (RaDaR™ assays) were designed for plasma sequencing.¹⁸ This assay can detect tumor VAF as low as 0.002%, with a positive estimated VAF was 0.042%.^18,19 In addition, large panels have been used for plasma sequencing. In a prospective study enrolled 91 patients with colorectal cancer liver metastasis (CRLM), primary (n = 50) and metastatic (n = 48) tissues and 271 serial plasma samples were used for molecular analysis.²⁰ Tumor samples were sequenced using WES, with an average sequencing depth of 500 ×, plasma samples were sequenced using a personalized panel of 451 genes, with an average sequencing depth of 20,000 ×. MRD positivity was defined as a VAF greater than 0.5%. In this particular instance, MRD detection may serve as a potential solution by combining the strengths of tumor-informed assays and tumor-agnostic assays, thereby warranting further discussion as a means to overcome the limitations associated with both approaches.

According to the above studies, although WES was performed for tissue sequencing, the increase number of variants in personalized panel for plasma sequencing is accompanied by the reduce in sequencing depth and detection sensitivity. However, tumor-informed analysis still showed a relative high sensitivity, as it can detect variants with allele frequency (VAF) of less than 0.01%.¹⁹ A massively parallel sequencing technique called SafeSeqS was used for tumor-informed assay to detect low frequency mutation. It was found that recurrence risk increased exponentially with increasing ctDNA mutant allele frequency (MAF) (hazard ratio, 1.2, 2.5 and 5.8 for MAF of 0.1%, 0.5% and 1%).²¹ Furthermore, the longitudinal sensitivity and specificity of MRD (88% and 98%, respectively) for recurrence prediction were much higher than those of CEA (69% and 64%, respectively).¹⁵ This indicated the high sensitivity of ctDNA for MRD monitoring.

However, the tumor-informed assay isn’t perfect. It usually starts with a tissue WES assay or gene panels to obtain the personalized mutation information. So, for patients with difficulty to get tumor tissue, this method would not applicable. Besides, it may take a longer turnaround time, which may affect the decision making on adjuvant therapy. In addition, because of the heterogeneity of tumors that may undergo clonal mutations during treatment or patients who may have more than one primary malignancy, tumor-informed strategies are unable to detect new secondary primary tumors and false negatives may occur.

2.2. Tumor-agnostic assays

Tumor-agnostic assays, also called tumor-uninformed assays or tumor-naïve assays, as this method would not get genomic information from primary tumor tissue. ctDNA detection and analysis are carried out by only a fixed panel, which is designed by a group of pre-selected primers or probes. In a study, a panel including 338-gene was performed to detect MRD. This study enrolled 261 stage I – III NSCLC patients who underwent definitive surgery, and 913 peripheral blood samples were collected. The median sequencing depth was 1214 ×, with a limit of detection achieved 0.1%, and a sensitivity of 97.9% for SNV or small indel detection. Positive MRD was defined as the detection of at least one somatic mutation.²² In another prospective clinical trial, 202-gene panel was used to detect ctDNA in blood samples from 20 patients with hepatocellular carcinoma. The median depth was 35,000 ×, and the limit of detection achieved 0.1%. The positive MRD was defined as the detection of at least one somatic mutation.²³ Significantly differences in sequencing depth and LOD were shown in these studies. Furthermore, the LOD may be lower than tumor-informed methods.

In recent year, multi-omics methods, such as ctDNA methylation or fragment-omics were applied in tumor-agnostic assays to improve its performance, including specificity, sensitivity, sequencing depth and LOD²⁴. For example, a prospective study analyzed plasma samples from 103 colorectal cancer patients who underwent radical surgery and correlated them with recurrence. At the landmark plasma draw, ctDNA was detected in 15 patients, and all 15 patients relapsed [positive predictive value (PPV), 100%; HR, 11.28 (P < .0001)]. The characteristic recurrence sensitivity and specificity were 55.6% and 100%, respectively. Compared with tumor-naive somatic alterations alone, this approach increased sensitivity, with longitudinal sensitivity and specificity for recurrence approaching 69% and 91%, respectively.²⁵ A representative of the immobilized panel of tumor uninformed assay is cancer personalized profiling by deep sequencing (CAPP-Seq). In another study collected 77 plasma samples from 30 patients with stage I – IVA human papillomavirus-negative HNSCC and plasma samples from 20 risk-matched healthy controls for personalized analysis of CAPP-Seq and cell-free methylated DNA. Personalized analysis of methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq) were used to detect ctDNA-derived somatic mutations and aberrant methylation, respectively. Candidate mutations within the lower 15th or upper 85th percentile of the sequencing depth (≤1,500X and ≥ 5,000X) of HNSCC cfDNA or PBL gDNA samples were included for analysis.²⁶

Taken together, the tumor-agnostic approaches are usually designed with a fixed panel and are now often combined with multiomics methods. Therefore, the turnaround time and cost are less with the tumor-agnostic strategy than with the tumor-informed strategy. Although this method facilitates large-scale screening of suspicious genes, but this advantage also entails a reduction in test specificity and an increase in false-positive results.

2.3. Comparison of the advantages and disadvantages between tumor-informed and tumor-agnostic assays

With the increased clinical availability of ctDNA MRD, there is a growing number of commercially available testing products, including tumor-informed and tumor-agnostic assays (Table 1). The turnaround time and cost are less with the tumor-agnostic strategy than with the tumor-informed strategy; this may affect the decision making in adjuvant therapy and patient selection. In addition, because of the heterogeneity of tumors that may undergo clonal mutations during treatment or patients who may have more than one primary malignancy, tumor-informed strategies are unable to detect new secondary primary tumors and false negatives may occur. Therefore, tumor-informed assay is more suitable for recurrence prediction of a single primary tumor, although it may yield false-positive cases that require histopathologic confirmation of the tumor. For tumor-informed ctDNA assays, the longitudinal sensitivity and specificity for recurrence were reported to be close to 88% and 98%, respectively; in the same cohort, these were much higher, compared with those of CEA (69% and 64%, respectively).¹⁵ In a recent study on uninformed tumor plasma ctDNA combination, the longitudinal sensitivity and specificity for recurrence were close to 69% and 91%, respectively.²⁵ The difference in analytical sensitivity between these assays may be attributed to inherent differences in the tumor informing the top hits of NGS assays, which are then used for plasma ctDNA testing. Taken together, the tumor-agnostic approach facilitates large-scale screening of suspicious genes, but this advantage also entails a reduction in test specificity and an increase in false-positive results. However, the cost of multiplex testing, false-negative results, tumor tissue accessibility, and longer turnaround times are potential barriers to the use of this approach. Currently, the choice of testing strategy may depend on the level of evidence from ongoing randomized clinical trials, investigator preference, cost-effectiveness, patient economics, and availability of tumor tissue.

Table 1.

Clinical trials of completed studies for the detection of ctDNA MRD.

	Enrolled patients	ctDNA detection method	Testing panel	Cutoff of ctDNA detection	Timing of ctDNA sampling	Results & Comments	Ref
1	261 (I-III) NSCLC accepted surgery who received surgery + adjuvant therapy (n =55) or surgery only (n =195)	Tumor-informed& Tumor Naïve	338 genes	For tissue-derived variants: ≥ 2 reads; For variants only in cfDNA: Hotspot: ≥4 reads; Non hotspot: ≥8 reads;	Landmark: within 1 month after definitive treatment	3.2% with longitudinal undetectable MRD recurred (NPV=96.8%); PPV of longitudinal detectable MRD was 89.1%, with a median lead time of 3.4 months; The peak risk of developing detectable MRD was approximately 18 months after landmark detection; Brain-only recurrence was less commonly detected by MRD (n = 1/5, 20%); Patients with undetectable MRD might not benefit from adjuvant therapy;	22
					Longitudinal: every 3-6 months
2	22(IB-IIIA) NSCLC accepted IO+Chemo (n=12) or IO+IO (n=4) or Chemo (n=6)	Tumor Naïve	194 genes	≥ 1 variant	3-8 d before NAT	RFS: 7.41 (0.91–60.22);	33
					3-8d after surgery	RFS: 5.37 (1.27–22.67)
3	88(IA-IIIB) NSCLC accepted surgery (N=61); surgery + adjuvant chemotherapy/ radiotherapy (N=8); chemoradiotherapy (N=19)	Tumor-informed	t: WES b:47 variants	≥ 1 variant	Baseline prior to treatment	OS:2.97 (p= 0.01); RFS: 3.14 (p=0.003)	18
					2 w to 4m following the end of treatment	RFS:14.8 (p<10-5); OS: 5.48 (p<0.0003)
4	65 stage IA-IIIB NSCLC	plasma-based mutation and tumor-informed methylation	168-genes Mutation and DNA methylation	MRD score	Baseline: before surgery	MRD score was observed between 0.5-7 months prior to radiologic recurrence for all 5 patients, but1 patient had no mutation detected throughout all follow-ups.	65
					Postoperative follow-up
5	116 NSCLC received surgery + adjuvant chemotherapy	Tumor Naïve	139-genes	≥ 1 variant	Before surgery	ctDNA positive patients had a significantly reduced RFS compared with ctDNA negative patients (HR: 4.0; 95% CI: 2.0–8.0; p < 0.001;	66
					After surgery 30 d
					follow-up (every3 m)
6	20 IIA-IIIA NSCLC received surgery or surgery + adjuvant chemotherapy	Tumor Naïve	197-genes	≥ 1 variant	24–48 h before surgery	Postoperative positivity for ctDNA also predicted shorter recurrence-free survival (RFS) (P=0.015)	67
					After surgery 3–12 d
					follow-up
7	36 NSCLC patients who accepted surgery and 15 patients with non-malignant disease	tumor-informed	197 genes; Variant and ctDNA fragment score;	At least 1 to 6 baseline-informed variants	Preoperative	The combination of variants and ctDNA fragment score would distinguish patients with disease recurrence with 75% accuracy (P = 0.0029); This integrated method may be cheaper and more sensitive.	68
					Postoperative
8	82 locally advanced rectal cancer received neoadjuvant + surgery	Tumor-informed & Tumor Naïve	t:509 genes panel or WES / b: 22 / 19 genes panel	≥ 0.05%	NAT:2 weeks after NAT	RFS: HR=27.38; P< 0.0001); OS: HR = 17.78(95% CI,1.94-162.60; P= 0.00054)	34
9	455 stage II CRC received surgery + adjuvant or surgery	Tumor-informed	t: Safe-SeqS (15 genes panel) / b: Safe-SeqS PCR	P < .1	Randomized trial of ctDNA-guided adjuvant chemotherapy:4 or 7 weeks after surgery	15% patients in the ctDNA-guided group received adjuvant chemotherapy vs. 28% in the standard-management group; 3year RFS was 86.4% ctDNA(positive) + adjuvant chemotherapy group vs. 92.5% in ctDNA negative + no adjuvant chemotherapy group	41
10	84 (stage I (9.5%), II (23.8%), III (47.6%), IV (19%)) CRC received surgery only (n=39) or completion of adjuvant therapy (n=45)	Tumor Naïve	b: Guardant Reveal® (500 kb panel)	/	four weeks post-completion of definitive therapy	Sensitivity of recurrence detection of integrating genomic and epigenomic signatures, epigenomic calls alone, and genomic calls alone was 55.6%, 48.2%, and 40.7%, respectively.	25
11	119 locally advanced CRC who received neoadjuvant chemoradiotherapy (nCRT)+ surgery	Tumor Naïve	b: 422 panel	/	post-neoadjuvant chemoradiotherapy	ctDNA clearance was associated with pCR; RFS: HR = 9.29, 95% CI = 3.74 to 23.10, P < 0.001;	35
					post-surgery	RFS: HR = 90.29, 95% CI = 17.01 to 479.26, P < 0.001
12	91 colorectal liver metastases patients who received surgery + adjuvant chemotherapy	Tumor-informed	t: WES/ b: 451 panel	VAF ≥ 0.5%	Post-surgery	The overall recurrence rate in ctDNA-negative patients was 41.7% and 79.4% in ctDNA-positive patient; RFS:HR 0.374; 95% CI, 0.205-0.682; P = 0.001	20
					post-ACT	The overall recurrence rate was 40.7% in ctDNA-negative patients and 77.3% in ctDNA-positive patients; RFS: HR 0.406; 95% CI, 0.189-0.873; P = 0.021
13	24 oligometastatic CRC who received neoadjuvant chemotherapy + surgery; or surgery	Tumor Naïve	b: Roche AVENIO Surveillance panel 197 panel	/	Pre-operation	Plasma ctDNA detection with 95% sensitivity and 100% specificity;	69
14	142 Breast cancer (stage 0-III) who received hormonal therapy, adjuvant local and systemic therapy	Tumor-informed	t: WES b: NGS	≥2 variants	After surgery	Distant recurrence: HR=5.1 (95% CI, 2.0-12.7	70
					One year after surgery	Distant recurrence: HR=20.8 (95% CI, 7.3–58.9)
15	170 early-stage breast cancer patients received neoadjuvant chemotherapy + surgery (n = 140); or surgery + adjuvant chemotherapy (n = 30)	Tumor-informed	t: WES; b; dPCR	/	Diagnosis	RFS: HR=5.8 (95% CI, 1.2-27.1)	3
					Follow-up	ctDNA detection had a median lead time of 10.7 months (95% CI, 8.1-19.1 months) compared with clinical relapse
16	103 stage II-III, HR+, breast cancer patients	Tumor-informed	t: WES; NGS-36 panel	/			71,72
17	84 stage II-III and high-risk early breast cancer patients who received neoadjuvant chemorapy (NAC)+ surgery	Tumor-informed	t: WES; NGS-16 panel	≥ 2 variants	Before neoadjuvant (T0);	metastatic recurrence: HR = 4.1 (95% CI, 0.52-32.4)	54
					3 weeks after initiation of paclitaxel (T1)	metastatic recurrence: HR = 4.5 (95% CI, 1.16-17.42)
					Between paclitaxel and anthracycline regimens (T2)	all patients who achieved pCR were ctDNA negative; ctDNA-positive patients (14%) had a significantly increased risk of metastatic recurrence (HR = 10.4, 95% CI: 2.3-46.6
					Prior to surgery (T3)	non-pCR but were ctDNA negative (86%) had excellent outcome similar to those who achieved pCR (HR 1.4; 95% CI 0.15-13.5)
18	581 UC (ypT2–4a or ypN+; ypT2–4a or ypN+) patients who received surgery + immunotherapy	Tumor-informed	t: WES; b: NGS-16 panel	≥ 1 variant	The first day of immunotherapy	DFS: HR = 0.58 (95% CI, 0.43–0.79);	8
					The first day after immunotherapy of 6 weeks	ctDNA clearance was higher in the atezolizumab arm (18%) than in the observation arm (4%) (P = 0.0204)
19	68 advanced MIBC who accepted surgery + chemotherapy	Tumor-informed	t: WES; b: NGS-16 panel	≥2 variants	Before chemotherapy	ctDNA (+): 46% (11/24) recurrence; ctDNA (-): 3% (1/35) recurrence	2
					Before cystectomy	ctDNA (+): 75% (6/8) recurrence; ctDNA (-): 11% (6/55) recurrence; all ctDNA positive patients had residual tumor and/or lymph node metastases
					After cystectomy	ctDNA (+): 46% (11/24) recurrence; ctDNA (-): 3% (1/35) recurrence; 100% sensitivity, 98% specificity in monitoring metastatic relapse
20	42 localized MIBC proceeding radical cystectomy and 15 healthy adults	Tumor Naïve	Urine: uCAPP-Seq	VAF 2.3%	After cystectomy	PFS HR = 7.4 (95% CI, 1.4-38.9); sensitivity of 81% and specificity of 81% in monitoring recurrence	72

Note: NSCLC: nonsmall-cell lung cancer; CRC: colorectal cancer; UC: urothelial carcinoma; MIBC: muscle-invasive bladder cancer; NPV: negative predictive value; PPV: positive predictive value; t (tissue) and b (blood) in the panel column

3. Clinical application of MRD in cancer

3.1. Non small-cell lung cancer (NSCLC)

NSCLC is the leading cause of cancer-related death, and patients with disease recurrence showed an inferior survival. MRD would help to timely discover recurrence or monitor the response to treatment would help to optimize the treatment decisions. Herein, we reviewed the application of ctDNA MRD detection in NSCLC.

3.1.1. Predicting response and guiding therapy in NSCLC

Firstly, MRD detection helps predict response in patients with advanced lung cancer. In a real-world study, dynamically monitored ctDNA was shown to predict clinical treatment efficacy in NSCLC.²⁷ The study found that higher baseline ctDNA content or higher number of variants in plasma was associated with shorter overall survival (OS). In addition, patients with ctDNA clearance after treatment had relatively long progression-free survival (PFS) and OS, independent of treatment method and assessment time.

Secondly, MRD could predict the treatment response of lung cancer patients receiving targeted therapy or immunotherapy. For targeted therapy, the persistently positive ctDNA can predict the risk of early progression after EGFR-TKI therapy. In one study of patients treated with osimertinib plus bevacizumab, those with EGFR-mutant ctDNA after 6 weeks of treatment had earlier progression and worse OS.²⁸ For immunotherapy, in patients treated with immune checkpoint inhibitors, ctDNA response (i.e., >50% reduction) was associated with better PFS and OS and was consistent with imaging-assessed efficacy. Median OS was significantly prolonged in those who were responsive by ctDNA (31.2 months vs. 18.4 months; HR = 0.36).²⁹ In the prospective INSPIRE study, all 12 patients whose ctDNA was cleared during pembrolizumab treatment had favorable clinical outcomes. For patients who had ctDNA testing during the surveillance period, ctDNA levels below baseline showed a median OS of 23.8 months. However, the elevated ctDNA levels above baseline were associated with disease progression and worse survival, with a median OS of 13.7 months.¹⁷

Thirdly, the detection of MRD can provide reference value for adjuvant therapy of NSCLC patients. MRD-positive patients benefited more from adjuvant therapy than MRD-negative patients. Recently, Wu Yilong et al. (2022) found that the presence of MRD in 10 of 55 (18.2%) NSCLC patients who received adjuvant therapy, including chemotherapy, targeted therapy, and immunotherapy, had significantly improved disease-free survival (DFS). The remaining 45 patients who were MRD-negative could not benefit from adjuvant therapy (P = .022; HR = 0.34; 95%CI:0.12–0.88)²². Moreover, patients with negative postoperative ctDNA MRD testing had extremely low tumor burden, and these patients may not require adjuvant therapy. This study demonstrates that the identification of ctDNA in plasma can guide treatment decisions in NSCLC patients.

3.1.2. Predicting recurrence in NSCLC

MRD assessment by ctDNA analysis predicts recurrence after treatment of early-stage lung cancer with exceptional accuracy.^22,30,31 The TracerX study evaluated the predictive value of the MRD test for postoperative recurrence of NSCLC. Thirteen of 14 relapsed patients had measurable ctDNA prior to clinically apparent disease. The median interval between ctDNA detection and radiologic diagnosis was 70 days.³² Another study including 261 patients with stage I to III NSCLC who underwent radical surgery, were enrolled, and the prognosis of single lymph node MRD-negative patients was significantly better than that of MRD-positive patients (HR = 0.08; 95% CI: 0.02–0.33).²² Especially those with undetectable longitudinal MRD may represent a potentially curable population.

Furthermore, MRD detection plays a role in predicting recurrence in patients who received neoadjuvant therapy. A retrospective study of 22 patients with stage IB-IIIA NSCLC who received neoadjuvant therapy showed that the results of ctDNA dynamic analysis were highly consistent with postoperative pathologic evaluation; that patients with positive ctDNA at 3 to 8 days after surgery had relatively high recurrence risk (HR = 5.37); and that the molecular recurrence prediction based on ctDNA NGS detection was 6.83 months earlier, compared with the median time of assessment by imaging criteria.³³

Therefore, more studies with larger patient populations and multiple different post-treatment time points are needed to better understand ctDNA MRD and its predictive association with NSCLC recurrence and survival.

3.2. Colorectal cancer (CRC)

Colorectal cancer (CRC) is the third most common cancer and the fourth leading cause of cancer-related death worldwide. Recently, ctDNA monitoring was included in the latest edition of National Comprehensive Cancer Network (NCCN) colon cancer guidelines (2021.V2 edition) for CRC clinical whole process management. ctDNA dynamic monitoring enables refined management of patients by predicting recurrence, evaluating efficacy and risk stratification.

3.2.1. Predicting response and guiding therapy in CRC

Neoadjuvant chemoradiotherapy (nCRT) is one of the effective treatments for patients with locally advanced rectal cancer.^34,35 Currently, the commonly used clinical tools, such as magnetic resonance imaging, endoscopy and digital rectal examination (DRE), lack accuracy and timeliness in predicting patient response to nCRT.³⁵ Some prospective cohort studies have confirmed that ctDNA status after nCRT can predict nCRT efficacy.^34,35 In the study by Wang et al., the clearance of ctDNA during nCRT was found to have a significant correlation with the pCR status. importantly, detection of ctDNA-MRD identifies patients who have received nCRT and may have achieved a pathological complete response (pCR) before surgery. These patients may benefit from a “wait and see” approach rather than total mesenterectomy (TME) plus adjuvant therapy.³⁵ In another study, ctDNA of eight patients was monitored longitudinally, two patients who had complete ctDNA clearance during adjuvant chemotherapy (ACT) did not relapse, whereas all the remaining six patients who were ctDNA-positive during ACT or shortly after completion of ACT relapsed.¹⁵ So, ctDNA monitoring offers the opportunity to predict prognosis preoperatively and guide subsequent therapy.^36,37

In stage II CRC, the postoperative management is controversial.¹⁴ Positive ctDNA, along with other prognostic risk factors, may provide support for considering chemotherapy in high-risk stage II CRC patients who are MSS/pMMR. Ongoing prospective studies, such as COBRA and DYNAMIC-II, investigating the use of ctDNA in stage II CRC, will be useful in guiding the appropriate use of ctDNA for making management decision.^38,39 ctDNA MRD can identify high-risk patients and who need to receive adjuvant therapy in stage II CRC. For patients with stage III CRC, ctDNA-MRD can help guide de-escalation or escalation of chemotherapy and is a biomarker to guide adjuvant treatment decisions.^36,37,40,41 In a study.⁴¹ 302 patients were assigned to the ACT (ctDNA-positive) or without treatment (ctDNA-negative) groups, based on their ctDNA status at 7 days or 4 weeks after surgery; there was no significant difference in the three-year RFS rate between the ctDNA-positive and ctDNA-negative groups (86.4% vs. 92.5%). This ctDNA-guided clinical trial revealed that patients with stage II CRC may be screened according to ctDNA status in order to provide the precise therapy.

Therefore, ctDNA MRD dynamic monitoring can be used to predict and evaluate the efficacy of colorectal cancer, including neoadjuvant and postoperative adjuvant treatment. Importantly, more clinical trials designed based on ctDNA biomarkers are being carried out, aiming to study the feasibility of adjusting the clinical treatment regime according to the status of ctDNA. The TRACC research is an ongoing clinical trial on high-risk stage II and stage III CRC, with dosing regimens adjusted according to the ctDNA status.³⁷ Its primary endpoint was designed to demonstrate noninferiority in the three-year DFS between the standard of care and ctDNA-guided ACT arms.

3.2.2. Predicting recurrence

According to the NCCN guidelines, neoadjuvant chemoradiation (nCRT) combined with total mesorectal excision (TME) is the standard treatment for locally advanced rectal cancer (LARC) standard treatment.⁴² ctDNA monitoring before and after nCRT provides the opportunity to predict prognosis preoperatively and guide subsequent therapy.^34,35 The results of Wenyang Liu et al. showed that ctDNA-positive (MRD-positive) patients after neoadjuvant therapy had a significantly higher risk of recurrence than MRD-negative patients (HR = 27.38; 95% CI, 8.61–87.06; log-rank P < .0001), with shorter OS and recurrence-free survival (RFS).³⁴ In the study, 17 of 60 patients with locally advanced colorectal cancer (LARC) had recurrence, of which 76.47% (13/17) were ctDNA-positive in the post-nCRT blood samples with a sensitivity of 76.47% and a specificity of 97.67% in recurrence prediction.

The postoperative presence of MRD in CRC is associated with a high risk of recurrence. Ambulatory monitoring of ctDNA identified MRD after radical surgical resection and prompted suspicion of recurrence 16.5 months earlier than clinical assessment alone.¹⁵ The NCCN guidelines recommend postoperative ctDNA as a marker for increased risk of recurrence in stage I – III CRC.⁴² Postoperatively, ctDNA-positive patients are more likely to relapse and have shorter RFS compared with ctDNA-negative patients.^{15,20,41,43,44} In a prospective study, ctDNA analysis was performed on 829 blood samples from 125 stage I – III CRC patients;¹⁵ ctDNA status on postoperative day 30 was significantly associated with recurrence risk, ctDNA-positive patients were seven times more likely to relapse compared with ctDNA-negative patients (HR = 7.2; 95%CI:2.7–19.0; P < .001).

In addition, studies have shown that the ctDNA status of postoperative adjuvant therapy was significantly associated with relapse risk.^{14,15,20,24,37,41,43,45} In a prospective study of 91 patients with CRC who underwent resection of liver metastases, ctDNA was analyzed in 271 blood samples and 83 patients underwent postoperative ACT.²⁰ The overall recurrence rates for ctDNA-positive and ctDNA-negative patients after ACT were 77.3% (17 of 22) and 40.7% (11 of 27), respectively, indicating that ctDNA-positive patients were more likely to relapse after ACT. Compared with ctDNA-positive patients, ctDNA-negative patients had significantly prolonged RFS after ACT (HR = 0.406; 95%CI:0.189–0.873; P = .021).

Therefore, studies have shown that ctDNA status during perioperative period can predict the recurrence of patients, and ctDNA-positive patients have a higher risk of recurrence compared with ctDNA-negative patients.

3.3. Genitourinary cancer

In patients who underwent radical resection of non-muscle invasive bladder cancer (NMIBC), the one-year recurrence rate was as high as 60% and there was progression to muscle invasive bladder cancer (MIBC). Therefore, regular follow-up is necessary^46,47, and MRD showed its advantages in dynamic monitoring.

3.3.1. Predicting response and guiding therapy in genitourinary cancer

ctDNA can be used to monitor response to treatment, and its clearance correlates with good prognosis. In a study, pembrolizumab were combined with stereotactic body radiotherapy (SBRT) for metastatic uroepithelial carcinoma (mUC), blood was collected for ctDNA analysis at baseline, during treatment and at the end of treatment. A rapid decrease in ctDNA was observed in responders compared to a stable or increasing status of ctDNA in non-responders⁹. In a phase III trial, the postoperative adjuvant efficacy of atezolizumab in patients with operable uroepithelial carcinoma was investigated. Patients who switched from positive to negative ctDNA MRD within two cycles of dosing showed longer DFS (HR = 0.26; 95%CI:0.12–0.56) and OS (HR = 0.14; 95%CI:0.03–0.59). These suggested that the clearing of ctDNA could predict better clinical benefit.⁸

Futuremore, MRD can be used as a stratification indicator to screen patients who can benefit from treatment. In perioperative chemotherapy studies in UC, patients with ctDNA negative or ≤ 2%, regardless of pT or pN staging, were considered by the investigators to be exempt from neoadjuvant chemotherapy.⁴⁸ In uroepithelial cancer, tumor DNA is also released in urine, and mutations in urine are highly consistent with tumor tissue, and these urinary tumor DNA (utDNA) can be used to guide neoadjuvant therapy.⁴⁹ In a study, MIBC patients undergoing sequential surgery with neoadjuvant chemotherapy were enrolled, baseline utDNA detected a total of 52 non-silent and 17 silent mutations, and the pathological response of postoperative tissues was significantly correlated with non-silent mutant utDNA. The sensitivity and specificity of positive utDNA with non-silent mutations for non-pCR was 81% and 81% (AUC = 0.78; 95% CI: 0.62–0.93), but the sensitivity of positive utDNA with silent mutations only reached 12%.⁵⁰ This study illustrates that utDNA-MRD may help screen patients who may benefit from neoadjuvant therapy.

In addition to the neoadjuvant phase, MRD has shown potential to guide treatment during the adjuvant phase. The IMvigor010 study proved that MRD-positive UC patients would enjoy better DFS (HR = 0.58; 95%CI:0.43–0.79) and OS (HR = 0.59; 95%CI:0.41–0.86) if accepted postoperative adjuvant therapy, however, in ctDNA-negative patients, the therapy would not change the DFS (HR = 1.14; 95%CI:0.81–1.62) or OS (HR = 1.31; 95%CI:0.77–2.23), so MRD can screen patients who can benefit from adjuvant therapy and reduce overtreatment.⁸

3.3.2. Detecting recurrence

A positive postoperative ctDNA MRD test is direct evidence of the presence of occult cancer cells. After cystectomy, ctDNA MRD monitoring correctly identified all patients with metastatic recurrence (100% sensitivity, 98% specificity) during disease surveillance and for up to 245 days (median 96 days), compared with the imaging lead time.² Similar finding was obtained in a prostate cancer study, which showed that baseline ctDNA abundance was an independent prognostic factor that correlated with shorter time to progression (HR = 2.38, P < .001) and OS (HR = 38.22, P < .001).⁵¹ Furthermore, plasma tumor fraction calculated from ctDNA was found to have a prognostic effect on OS (HR 1.75; 95%CI:1.08–2.85; P = .024), thereby, providing additional evidence for clinical treatment.⁵²

3.4. Breast cancer

3.4.1. Efficacy evaluation

Longitudinal analysis of ctDNA MRD shows promise for monitoring treatment response. Breast cancer patients who achieved pCR had significantly reduced ctDNA during neoadjuvant therapy, demonstrating the high accuracy of ctDNA analysis in assessing molecular responses during neoadjuvant therapy.⁵³ Magbanua et al. found that all patients who achieved pCR during neoadjuvant treatment were MRD-negative. Among patients who did not achieve pCR, those who were ctDNA-positive had a significantly increased risk of metastatic recurrence (HR = 10.4; 95% CI:2.3–46.6), but those who were ctDNA-negative had similar excellent outcomes as those who achieved pCR (HR = 1.4; 95%CI:0.15–13.5).⁵⁴ These studies suggested that personalized monitoring of ctDNA during neoadjuvant treatment for breast cancer may help to assess treatment response in real time and provide information on the modality and extent of subsequent treatment.

In patients with metastatic breast cancer, the molecular tumor compliance index of ctDNA may also serve as a potential biomarker of treatment response, with a concordance rate of 88.5% for disease remission and 87.5% for disease progression, compared with CT assessment results.⁵⁵ Clinical studies with larger samples are needed in the future to validate ctDNA MRD assessment of therapeutic efficacy and to refine clinically relevant diagnostic thresholds.

3.4.2. Predicting recurrence

Breast cancer remains one of the leading causes of cancer-related deaths in women worldwide, and some early-stage breast cancer develop recurrence and metastasis within 5 years of primary tumor resection.⁵⁶ Therefore, early detection of metastatic recurrence is important in clinical. In patients with triple-negative breast cancer treated with neoadjuvant chemotherapy, the risk of recurrence was higher in patients with post-surgical ctDNA-positive status than in patients with undetectable ctDNA (HR = 2.65; 95% CI:0.74–9.44). In addition, positive ctDNA MRD was associated with relapse of all subtypes of breast cancer, and its median time of detection was 7.9–10.7 months earlier than the clinical signs of tumor relapse; combined with the high clinical validity of molecular recurrence testing, MRD testing can be applied to early intervention in follow-up treatment in the future.^3,7 On the other hand, studies showed that positive ctDNA MRD before neoadjuvant therapy were significantly associated with worse DFS and OS.^57,58 Besides, the positive ctDNA after neoadjuvant therapy, was not only associated with worse DFS (HR = 2.67; 95%CI:1.28–5.57) and OS (HR = 4.16; 95%CI:1.66–10.42), but also predicted the higher risk of recurrence and worse event-free survival.^54,59

Therefore, ctDNA MRD detection of recurrence has a high clinical validity. In future, clinical trials would explore whether in cases of postoperative ctDNA MRD detection, treatment is needed immediately rather than waiting until clinical recurrence.

4. Challenges of the ctDNA-based MRD detection

4.1. False-negativity rate

A major issue facing MRD testing in solid tumors is the analytical sensitivity (i.e., false-negative rate) of ctDNA testing. First, the very limited total number of genome copies in blood samples limits the number of target-specific variants and can be demanding, when the VAF is reduced to near the LOD. Second, some biological properties of early-stage tumors may not favor ctDNA release and could present additional challenges for early detection efforts.^22,30,60,61 The third is the anatomical location of occult micrometastatic disease. In a study on 485 patients with nonmetastatic CRC, false-negative ctDNA testing at the immediate postoperative time point was more common in patients who subsequently experienced local recurrence alone.⁶²

4.2. False-positivity rate

Hematopoiesis from clones of indeterminate potential can lead to false-positive ctDNA results, which can be achieved by utilizing advanced bioinformatics analysis or by sequencing ctDNA with leukocytes and/or matched tumor tissue.¹¹ DNA fragments from non-neoplastic hematopoietic stem cells may lead to false-positive ctDNA results, especially in older adults. Somatic driver mutations detected in plasma likely originate from hematopoietic progenitor cells.⁶³

4.3. The optimal timing of blood collection after surgery to detect MRD

The optimal timing of postoperative blood collection for MRD testing needs to be determined in order to obtain more accurate information for adjunctive treatment decisions. For example, studies have shown that total cfDNA levels remain elevated for up to 4 weeks after CRC surgery; therefore, blood samples have higher rates of ctDNA detection when collected at 4 weeks after surgery than at earlier time points.⁶⁴ In addition MRD is a process of continuous and dynamic changes, and patients should be continuously monitored dynamically rather than relying on a single test result. Patients who are initially negative for ctDNA after surgery may later develop recurrence or metastasis.^3,15,22

5. Conclusions

ctDNA MRD has shown promising values in recurrence prediction and efficacy assessment in solid tumors cancers including lung, colorectal, urinary system, and breast. However, there are inevitable huge challenges to be overcome for routine clinical application. Firstly, the low sensitivity of ctDNA MRD monitoring for gene fusions and copy number abnormalities must be fully considered in clinical applications. Secondly, there is no standard on the sequence depth, and threshold of MRD positive etc. In future, the development of new technologies and the increase of clinical evidence would promote the application of MRD in clinical.

Biographies

Qiantong Dong, graduated from Wenzhou medical university with a medical doctor degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Chenbin Chen, graduated from Wenzhou medical university with a medical master degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Yuanbo Hu, graduated from Wenzhou medical university with a medical master degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Weiteng Zhang, graduated from Wenzhou medical university with a medical master degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Xinxin Yang, graduated from Wenzhou medical university with a medical master degree and is currently working at the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University.

Yingxue Qi, graduated from Shandong University of Traditional Chinese Medicine with a master degree and is currently working at the Jiangsu Simcere Diagnostics Co., Ltd.

Chan Zhu, graduated from Shanghai Ocean University with a master degree and is currently working at the Jiangsu Simcere Diagnostics Co., Ltd.

Xiaodong Chen, graduated from Wenzhou medical university with a medical doctor degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Xian Shen, graduated from the second military medical university with a medical doctor degree and is currently working at the First Affiliated Hospital of Wenzhou Medical University.

Weiping Ji, graduated from the second military medical university with a medical doctor degree and is currently working at the Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University.

Funding Statement

This study was supported by the National Natural Science Foundation of China (No.32070151); Key research and development project of Zhejiang Province (No.2021C03120); Key research and development project of Wenzhou City (No. ZY2021003); Natural Science Foundation of Zhejiang Province (No. LTGY23H030003); Wenzhou Science and Technology Project (No. Y20210174).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Authors’ contributions

Conception and design: Qiantong Dong, Xiaodong Chen, Xian Shen, Weiping Ji.

Acquisition of data (download publications, tables and figures): Yingxue Qi, Chan Zhu.

Writing, review, and/or revision of the manuscript: Qiantong Dong, Chenbin Chen, Yuanbo Hu, Xiaodong Chen, Weiteng Zhang, Xinxin Yang, Weiping Ji.