Liquid Biopsy in Advanced Colorectal Cancer: Clinical Applications of Different Analytes

Abstract

Colorectal cancer is one of the most prevalent cancers nowadays. In the metastatic setting, diagnosis and treatment have relied on tumor tissue analysis. However, the different limitations of this approach have recently opened the door to the introduction of liquid biopsy in the clinical setting. Liquid biopsy provides real-time information about the tumor and its heterogeneity in a simple, non-invasive, and repeatable way. There are several analytes that can be sought: exosomes, circulating tumor cells, and circulating tumor DNA, showing promising results in the areas of early detection, minimal residual disease, prognosis, or response to treatment. Here, we review the clinical applications of liquid biopsy in advanced colorectal cancer patients, focusing on metastatic diagnosis, prognostic assessment, drug sensitivity, treatment response, and acquired resistance monitoring.

1. Introduction

Colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancer-related death worldwide, with 1.1 million new cases per year [1]. About 20% of patients present an advanced stage at the time of diagnosis, and about 50% of those with localized disease will develop metastases over time [2]. The overall survival (OS) at 5-year in patients with stage IV disease is estimated to be around 15%, with a median overall survival (mOS) of approximately 30–40 months [3]. To date, the treatment algorithm of advanced CRC patients is based on performance status and molecular features assessed on surgical samples or tumor tissue biopsy. In particular, the mutational status of the Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS), Neuroblastoma Rat Sarcoma Viral Oncogene Homolog (NRAS), v-Raf Murine Sarcoma Viral Oncogene Homolog B (BRAF), and Human Epidermal Growth Factor Receptor 2 (ERBB2) genes, together with the Microsatellite Instability (MSI) status, are routinely assessed [2,4]. In recent years, new molecular biomarkers, such as Tumor Mutational Burden (TMB) and Neurotrophic Tyrosine Receptor Kinase (NTRK) 1–3 translocations, have emerged [5]. Intratumoral heterogeneity and emerging genomic changes due to disease spread and treatments limit tissue sample reliability in providing real-time biological disease information. In addition, tissue samples may be degraded or may be difficult to repeat during disease history due to patient compliance or a lack of technical feasibility. In order to overcome the different limitations of tissue biopsy, liquid biopsy is becoming increasingly important in clinical practice. Indeed, this is a non-invasive and repeatable approach, making it possible to provide a real-time picture of tumor heterogeneity to clinicians. Liquid biopsy can be performed from numerous body fluids, such as blood, saliva, stool, urine, cephalorachidian fluid, pleural fluid, or peritoneal fluid. Several analytes can be sought using this strategy; in particular, it is possible to search for Circulating Tumor Cells (CTCs), exosomes, and Circulating Tumor DNA (ctDNA). Each of these markers has been shown to play a role in early diagnosis, prognosis prediction, recurrence monitoring, treatment response assessment, and precision treatment guidance. Based on the prospective PRESEPT trial, the Food and Drug Administration (FDA) approved methylated Septin 9 (SEPT9) DNA detection by ctDNA in the CRC screening setting, with a sensitivity of 50% and specificity of 90% [6]. The overexpression of miR-17a and miR-130a analyzed at the level of exosomes has been shown to play a role in the early detection and prognosis of CRC [7]. In the case of early/localized disease and neoadjuvant therapies, thymidylate synthase+/RAD23 homolog B+ (Tyms+/RADS23B+) CTCs were found in up to 83% of patients without a clinicopathological response after neoadjuvant treatment; however, these CTCs were not detectable at baseline in patients with a complete or partial pathological response [8]. ctDNA has been shown to predict the risk of disease recurrence in both stage II and stage III patients [9,10]. Recently, Tie et al. conducted a prospective study in radically operated stage II patients, showing that a ctDNA-guided approach reduced the use of adjuvant chemotherapy without compromising the risk of recurrence compared with the standard strategy based on clinico-pathological high-risk characteristics [11]. In the metastatic setting, liquid biopsy may have important applications, both as a diagnostic and prognostic tool. Moreover, this strategy plays a role in the assessment of the treatment response, being able to predict it more effectively than known serum tumor markers, such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA 19-9). The role of liquid biopsy in metastatic CRC patients is not only limited to the evaluation of the response, thus leading to the early discontinuation of drugs that may be considered non-functional, but also in the assessment of primary and acquired resistance mutations that may direct the choice of first-line or subsequent-line treatment. In this context, there are several lines of evidence confirming a high concordance between the molecular result obtained by liquid and tissue biopsy in metastatic CRC patients across different treatment lines. Furthermore, Parikh et al. showed, in a prospective cohort of 42 patients with gastrointestinal cancers and acquired resistance to targeted therapies, that circulating free DNA (cfDNA) better identifies clinically relevant resistance mutations and other multiple mechanisms of resistance in 78% of cases than single-tissue biopsy [12].

In our review, we describe the potential roles of different liquid biopsy analytes (CTCs, exosomes, and ctDNA) in the management of metastatic colorectal cancer (mCRC) patients, analyzing their methodology and limitations, but focusing mainly on their clinical applications (Figure 1 and Figure 2).

2. Exosomes

2.1. Introduction, Methodology, and Limitations

Exosomes are extracellular vesicles of nanometric size (50–130 nm) enclosed in a lipid bilayer that protects them from degradation and makes them stable [13]. These small vesicles are secreted by various cell types and are present in almost all body fluids (blood, urine, cerebrospinal fluid, breast milk, saliva, etc.) [14]. Exosomes contain a variable spectrum of biologically active molecules, including nucleic acids (DNA, mRNA, microRNA, and long non-coding RNA), proteins, and lipids, reflecting the composition of the cells from which they originate, including cancer cells [13,15]. Several studies have shown that exosomes, by mediating communication between cells, play a role in tumor growth processes, metastatic niche formation, and immune system evasion [16]. The expression of proteins involved in altered cellular metabolic processes has been widely reported in exosomes isolated from patients with CRC [17]. In several studies, the concordance between molecular biomarkers detected in exosomes and tumor tissues has also been demonstrated [18,19,20]. These characteristics make exosomes very interesting analytes in the field of liquid biopsy.

Exosomes can be isolated by various methods, such as ultracentrifugation, density gradient separation, and antibody-based immunoaffinity capture, and it is also possible to analyze the expression levels of intra-exosomal non-coding RNAs [21,22]. Among these methods, ultracentrifugation is the gold standard for the isolation of exosomes, but requires a large amount of starting material, expensive equipment, and a lot of work. Commercial kits are available, but they are not standardized and involve the use of buffers that digest exosomal membranes, limiting the quantity and quality of analyses. Ultracentrifugation can also lead to the formation of particle clusters, limiting the characterization of individual vesicles [23]. Furthermore, their composition, but especially their small size, makes it difficult to develop a standard quantification approach [24].

2.2. Diagnosis, Prognosis, and Therapeutic Response Evaluation

Exosomes are being extensively researched for their possible role in the clinical management of mCRC patients, particularly for non-invasive diagnosis, tumor heterogeneity assessment, prognosis estimation, and therapeutic response monitoring (

Table 1. Summary of the biological functions and detection modes of nucleic acids and exosomal proteins in metastatic colorectal cancer patients.

Exosomal	Molecule Type	Sample Type	Detection Technique	Trend	Application
QSOX1	protein	plasma	ultracentrifugation	downregulated	diagnostic
Copine3	protein	plasma	ELISA	upregulated	diagnostic prognostic
miR-21	miRNA	plasma	ultracentrifugation	upregulated	diagnostic prognostic
miR-122	miRNA	serum	Invitrogen Exosome Isolation Kit	upregulated	diagnostic prognostic
miR-25-3p	miRNA	serum	ultracentrifugation	upregulated	prognostic
miR-28-3p let-7e-5p miR-106a-5p miR-542-5p	miRNA	plasma	TaqMan Low-Density Array	upregulated	diagnostic
KRAS mut	ExoDNA	plasma	ultracentrifugation ddPCR	upregulated	diagnostic prognostic predictive
UCA1	lncRNAs	serum	ultracentrifugation ExoQuick	downregulated	diagnostic
CRNDE-h	lncRNAs	serum	ExoQuick	upregulated	diagnostic prognostic
miR-17-92a	miRNA	serum	ultracentrifugation	upregulated	prognostic
miR-320d	miRNA	serum	ultracentrifugation	upregulated	diagnostic
miR-203	miRNA	serum	ultracentrifugation	upregulated	prognostic
miR-17-5p miR-92a-3p	miRNA	serum	ultracentrifugation	upregulated	diagnostic
hsa-circ0004771	circRNA	serum	Invitrogen Exosome Isolation Kit	upregulated	diagnostic prognostic
miR-196b-5p	miRNA	serum	exoRNeasy Kit	upregulated	predictive (5-fluorouracil resistance)
miR-21-5p miR-1246 miR-1229-5p miR-96-5p	miRNA	serum	ultracentrifugation	upregulated	predictive (oxaliplatin/5-fluorouracil resistance)
miR-222	miRNA	serum	ultracentrifugation	upregulated	diagnostic prognostic
UCA1	lncRNAs	serum	ultracentrifugation ExoQuick	Upregulated	predictive (cetuximab resistance)

Abbreviations: ELISA: Enzyme-Linked ImmunoSorbent Assay; miRNA: microRNA; ddPCR: droplet-digital polymerase chain reaction; ExoDNA: exosome DNA; lncRNAs: long non-coding RNAs; circRNA: circular RNA; UCA1: urothelial cancer-associated lncRNA 1; CRNDE-h: colorectal neoplasia differentially expressed-h.

).

Exosomal surface proteins, such as Quiescin Sulfhydryl Oxidase 1 (QSOX1) [25], Glypican-1 [26], and Copine3 [26,27], have been identified as promising biomarkers for the non-invasive diagnosis of CRC, as they have been found at much higher levels in the exosomes of CRC patients than in those healthy subjects. Similarly, several exosomal microRNAs (miRNAs) have been found to be differentially expressed in CRC patients and identified as potential diagnostic biomarkers [28,29,30,31]. In a study conducted by Tsukamoto on 326 CRC patients, the expression levels of exosomal miR-21 were significantly correlated with those detected in tumor tissue samples [32]. Similar correlations were reported in subsequent studies for miR-122 [33] and miR-25-3p [34].

Silva et al. identified a miRNA signature for the diagnosis of CRC by analyzing more than 100 plasma samples obtained from subjects with either localized or metastatic CRC, as well as from non-oncological patients [30]. A classifier based on four miRNAs (miR-28-3p, let-7e-5p, miR-106a-5p, and miR-542-5p) was shown to discriminate cancer from non-cancer cases among the analyzed plasma samples. The overexpression of these miRNAs was confirmed by reverse-transcription quantitative real-time polymerase chain reaction (RT-qPCR) and the data were externally validated. The sensitivity, specificity, and accuracy of the model were maintained when applied to tumor tissue samples [30].

The expression of miR-122 could help to specifically differentiate patients with and without liver metastases [33], and the overexpression of miR-320d proved to be useful in differentiating metastatic from non-metastatic CRC cases [35]. An Italian group monitored, for the first time, the mutational status of KRAS in CRC patients at different stages using exosomal DNA [36]. The expression level of exosomes was correlated with the extent of the disease. The resection of the primary tumor was correlated with a reduction in the KRAS G12V/D copy number, while CEA expression and the presence of liver metastases were correlated with a higher copy number [36].

Long non-coding RNAs (lncRNAs) could also find application in the non-invasive diagnosis of CRC, as they have been reported to be differentially expressed in the exosomes of CRC patients [37]. For instance, urothelial cancer-associated lncRNA 1 (UCA1) was downregulated in CRC [38], whereas the differentially expressed lncRNA of colorectal neoplasm-h (CRNDE-h) was upregulated compared with normal controls. In particular, the CRNDE-h levels were significantly associated with metastatic disease (p = 0.003) [39], and the exosomal circular RNA hsa-circ0004771 was also correlated with the presence of distant metastases [40].

As reported by several studies, exosomes are involved in the progression of CRC due to their role in mediating intercellular cross talk [41,42]. The exosomal load in mCRC, therefore, has prognostic importance.

Chen and colleagues extracted exosomes from the serum of CRC patients and identified 36 upregulated proteins involved in pro-metastatic modulation processes of the tumor microenvironment and 22 downregulated proteins involved in cell growth and survival [43]. The study showed that exosomes in CRC patients played a central role in promoting tumor invasiveness [43].

Exosomal dipeptidyl peptidase IV (DPP4) was identified as a potent inducer of angiogenesis through the activation of the Small Mother Against Decapentaplegic (SMAD) signaling pathway [44]. When DPP4 was inhibited, tumor growth was suppressed in in vivo models. In a study on 5-fluorouracil (5-FU)-resistant CRC, exosomal DPP4 concentrations were proposed as a useful prognostic marker, as they were correlated with the metastatic process [44].

The overexpression of exosomal miRNAs has recently been correlated with tumor progression and the development of liver metastases in CRC. For instance, the expression of the miR-17-92a cluster was associated with CRC recurrence and poorer prognosis if miR-17-92a was expressed at higher levels [45].

miR-25-3p miRNA can be transferred from CRC cells to endothelial cells via exosomes, promoting vascular permeability and angiogenesis [34]. It has been reported that exosomal miR-25-3p facilitates the formation of liver and lung metastases in mouse models of CRC and that the expression level of exosomal miR-25-3p in the serum of patients with mCRC was significantly higher than that in patients without metastases [34].

Another molecule that has been identified as a mediator of signaling between tumor cells and monocytes in patients with mCRC is miR-203, which promotes the differentiation of monocytes into tumor-associated M2 macrophages and has been associated with the development of liver metastases [46]. miR-203 has also been established as an independent negative prognostic factor in patients with CRC [46].

Microarray analysis of miRNAs showed the upregulation of miR-21 in exosomes, primary tumor tissue, and liver metastases [32]. Significant correlations were demonstrated between the exosomal miR-21 levels, TNM staging, and prognosis of CRC patients. The expression level of miR-21 was identified as an independent prognostic factor for OS and disease-free survival (DFS) in patients with stage II or III CRC and for OS in stage IV patients [32].

Similar data were reported for the molecules miR-17-5p and miR-92a-3p [47].

High levels of miR-222 at 24 weeks after the start of bevacizumab treatment have been shown to be associated with shorter survival in patients with mCRC [48].

Several studies have been conducted to assess a possible correlation between exosomal load and chemotherapy resistance in patients with CRC [49,50]. Exosomes could be an important element in the assessment of the treatment response and could represent biomarkers of treatment sensitivity and resistance [51]. Cancer Stem Cells (CSCs) are intrinsically resistant to chemotherapy [52]. Exosome-mediated communication with stromal fibroblasts can induce the de-differentiation of tumor cells into stem cells, promoting chemoresistance in CRC [53]. The main regulator of fibroblast-mediated reprogramming is the exosomal Wingless-related Integration Site (Wnt); thus, interfering with exosomal Wnt signaling could result in chemoresistance reduction [53].

The activation of the Extracellular Signal-Regulated Kinase/Protein–Kinase B (ERK/AKT) pathway by exosomes derived from CRC-associated fibroblasts appears to promote oxaliplatin resistance [54]. miR-196b-5p stimulates the maintenance of stem cell properties and the resistance of CRC cells to 5-FU via activation of the Signal Transducer and Activator of Transcription (STAT3) signaling pathway [55].

Recently, a panel of exosomal serum miRNAs containing miR-21-5p, miR-1246, miR-1229-5p, and miR-96-5p was developed to effectively distinguish 5-FU chemotherapy-resistant CRC cell lines from the chemo-sensitive control group [56]. High plasma levels of exosomal UCA1 have been shown to correlate with an inferior response to cetuximab treatment [57].

In addition to diagnostics and treatment monitoring, exosomes can also be used for therapeutic purposes to efficiently deliver drugs into various types of cancer cells, acting as vectors with low immunogenicity and high biocompatibility. Liang and colleagues used engineered exosomes to simultaneously deliver 5-FU and a miR-21 inhibitor oligonucleotide (miR-21i) to a 5-FU-resistant CRC cell line [58]. The engineered co-delivery system was able to efficiently facilitate the cellular uptake of the delivered molecules, reversing chemoresistance and significantly lowering miR-21 expression in cell lines and ultimately reducing tumor growth [58]. In another recent study, exosomes isolated from A33-positive cells were loaded with doxorubicin, generating an A33 antibody complex (A33Ab-US-Exo/Dox) aiming to target A33-positive CRC cell lines [59]. The A33Ab-US-Exo/Dox complex was shown to have excellent tumor-targeting and growth-inhibiting capacity for the target cancer cells [59].

Exosomes, therefore, perform a wide range of biological functions by transporting signaling molecules that regulate various cellular processes. Exosomes also contribute to CRC development and tumor invasiveness. Their easy extraction from a wide variety of biological fluids makes exosomes an attractive means to detect diagnostic, predictive, and prognostic information in mCRC patients. However, although preclinical data seem very promising, large studies are needed to assess the repeatability of the method and the applicability of exosomes as biomarkers for clinical management and therapeutic decisions.

3. Circulating Tumor Cells

3.1. Introduction, Methodology, and Limitations

CTCs are epithelial cancer cells that shed from the primary tumor and, after extravasation from the blood vessels through the epithelial–mesenchymal transition (EMT), acquire migration capacities and thus are considered one of the primary triggers of the metastasis process [60,61,62]. CTCs carry information from their primary tumor and newly gained phenotypic and genotypic characteristics that confer survival capacity in the circulating microenvironment that protects them from the immune system recognition and attack. These changes also provide them with interaction capacities with adhesion molecules and epithelial cells, as well as secondary metastatic niche formation potential [61,62]. CTC dynamics are also driven by their interaction with circulating proteins and growth factors, such as Fibroblast Growth Factor (FGF) and Vascular Endothelial Growth Factor (VEGF), generated by macrophages in the tumor microenvironment [62,63].

Conversely, the role of other immune cells and circulating components in CTC physiology has been described. Platelets play a fundamental role in CTCs’ protection and survival, helping them to evade cancer-associated fibroblast attack and promoting neutrophil recruitment at tumor sites, functioning as a shield against immunity [61,62,64]. Likewise, platelets can produce transforming growth factor beta (TGF-β), which facilitates EMT and vascular permeability [62,65]. In the case of neutrophils, their interaction with CTCs occurs through the upregulation of cell adhesion proteins, such as cadherin, integrin, and glycoproteins, as well as the release of NETS (Neutrophil Extracellular Traps), which are complexes of histone DNA and proteins that capture and promote CTC intravasation [61,66].

Detection techniques generally require the recognition, isolation, and separation of CTCs from other components by physical and biological differentiation (cell density, size, deformability, and polarity), followed by subsequent enrichment steps [67]. Among such techniques, those based on immune-affinity positive cell selection allow the recognition of epithelial or tumor surface antigens, such as cytokeratin (CK) 18, 19, and 20, and Epithelial Cell Adhesion Molecules (EpCAMs, CEA, and ERBB2). There are also immune-affinity negative selection techniques, which exclude circulating non-CTC cells based on the detection of leukocyte surface antigens usually not present on CTC membranes (i.e., CD45 and CD61) [60]. To date, the CellSearch system has been one of the most used techniques for recognizing CTCs, which is based on a positive selection through epithelial cell adhesion molecule antibody (EpCAM-based antigen detection) recognition, immunostaining for cytokeratin and epithelial cell markers, and CD45 leukocyte marker cell exclusion [67,68,69]. Other techniques use the Polymerase Chain Reaction (PCR) for the detection of cancer-specific genes in the selected cells without prior enrichment [60].

The use of CTCs in clinical practice is limited by technical difficulties regarding the detection and characterization processes due to their low frequency and counts in peripheral blood samples (1–10 CTCs/mL of whole blood) compared with 10⁷–10⁸ in circulating leukocytes [67,70], which is mainly related to their elimination by the immune system. Other limitations regarding the isolation process from other circulating cellular components are mainly related to their short half-life and the downregulation of CTC-associated markers [37,71], such as those of the epithelial lineage (EpCAM) usually detected by standard screening platforms, in favor of a mesenchymal phenotype [60,62].

3.2. Metastatic Diagnosis, Prognosis, and Therapeutic Response Evaluation

Multiple studies have focused on the diagnostic, prognostic, and predictive potential of CTCs and their role in early and advanced diagnosis, as well as dynamic change monitoring in response to treatment. Particularly in the case of CRC, the quantification of CTCs has served as a differential marker for discerning between benign and malignant diagnoses. Tsai et al. conducted a study to assess the diagnostic role of CTCs. This prospective trial quantified CTCs in patients with premalignant lesions, as well as colon tumors at different stages, in comparison with clinical diagnostic tests (i.e., enteroscopy). The results showed 88% accuracy, with a false positive rate of only 3.3% in healthy controls, as well as a false negative rate of 16% in patients with malignant lesions [72]. Healthy patients (2.11 CTCs) could be distinguished from those with adenomas (5.83 CTCs; p = 0.0001) or CRC (16.99 CTCs; p = 0.0001); furthermore, the mean CTC count for patients with advanced CRC (III-IV stage) was significantly different from the mean CTC count for patients with early stage CRC (p = 0.015) and adenomas (p = 0.016) [72].

A prospective study by Cohen et al. investigated a potential relationship between CTCs and radiological imaging response parameters, and prognosis in terms of Progression-Free Survival (PFS) and OS in mCRC patients, with a CTC cutoff of ≥3 CTC/7 mL. The results showed that CTC counts were significantly related to the disease’s aggressiveness in terms of poor clinical–radiological prognosis. Indeed 20% of CRC patients with radiological progression and 27% of deceased patients had unfavorable CTC counts in both study cutoffs (3 and 5 weeks after treatment start), compared with only 7% of the not-progressing subgroup [73]. The study also reported a sensitivity and specificity of 27% and 93%, respectively, in terms of concordance with radiological findings. Likewise, predictive and prognostic relationships were found, with worse median Progressive-Free Survival (mPFS) and mOS in both patients with poor baseline CTCs (mPFS 4.5 months vs. 7.9 months; p = 0.0002; mOS 9.4 months vs. 18.5 months; p < 0.0001) and those with persistently elevated CTCs throughout treatment [73]. Another study by Camera et al., focusing on the dynamic evolution of biomarkers (cfDNA, CEA, and CTC), as well as radiological changes as prognostic factors of OS in mCRC patients, failed to show a significant association with outcomes, while only baseline elevation of such biomarkers, especially CTC, predicted worse OS (HR 6.5) [37,74].

Conversely, CTC counts and their characterization can help to predict disease recurrences in CRC patients. In particular, CTC counts before and after tumor resection can predict the risk of metastasis, with both baseline and one-month-after high CTC loads anticipating poor DFS and OS in patients with metastatic, surgically resected liver metastasis [60,75]. Pan et al. also showed that CTC counts were correlated with tumor T, N, and M staging, with a directly proportional relationship between higher stage and CTC counts in both right- and left-side disease [76]. This correlation between higher CTC counts and worse prognosis was also confirmed in a meta-analysis by Tan et al., taking into account 15 studies and >3000 patients, showing that the presence of CTCs was correlated with worse OS (HR 2.28) and PFS (HR 1.53). Conversely, the multivariate analysis in the same study showed that the worse OS and PFS were independent of CTC detection timepoints (baseline or along treatment) [77].

Concerning the detection of the KRAS mutation, Fabbri et al. identified a 50% concordance between tissue sample and CTC molecular analyses in patients with mCRC using a dielectrophoresis-based platform able to detect and sort pure CTCs [78].

The role of CTCs as a dynamic monitoring tool and treatment response predictor has also been largely investigated. The study by Delgado-Urena et al. showed a concordant decrease in CTC counts throughout treatment in patients with mCRC experiencing a favorable clinical response or at least stable disease by RECIST [79].

In addition, CTC counts may also play a role in treatment tailoring. In this context, the VISNU1-Trial investigated the potential role of poor CTC counts (≥3) as a biomarker guiding treatment decisions. After comparing triple versus double therapy (FOLFOXIRI vs. FOLFOX), the study found that the former was associated with better mPFS (12.4 months vs. 9.3 months; HR 0.64, 95% CI 0.49–0.82; p = 0.0006) in patients with poor baseline CTC counts; ECOG 1, RAS, and BRAF mutations with a CTC count of ≥20 were considered as independent negative prognostic factors in this patient cohort [80].

A prospective study conducted in mCRC KRAS Wild-Type (WT) patients, mostly (90%) receiving an anti-Epidermal Growth Factor Receptor (EGFR) antibody during systemic treatment, evaluated the response-predicting and prognostic role of CTCs. Patients who achieved favorable kinetics, with CTCs below the established median, had a better mPFS than those with unfavorable kinetics (14.7 months vs. 9.4 months; p = 0.02) [81].

Finally, the role of CTCs in predicting treatment resistance has been considered. In this way, it has been seen that, throughout treatment, CTCs express resistance traits derived from the overexpression of genes related to the mammalian Target of Rapamycin (mTOR) and PhosphatidylInositol 3-Kinase (PI3K)/AKT cascades, which are usually associated with cell growth, proliferation, and chemotherapy resistance [82]. Similarly, the overexpression of CTCs characterized by upregulated xenobiotic metabolism enzymes and enhanced cytochrome p450 activity, responsible for detoxifying and metabolizing drugs, justifies the lower effect and final resistance to chemotherapy (ChT) [82,83]. In this sense, a study by Grillet et al. showed a lower response to in vitro FOLFIRI ChT schemes in CTCs compared with primary tumor cells (p = 0.0109), which was likely related to higher expression of irinotecan-resistance genes, such as UDP-glucuronosyltransferase 1-1 (UGT1A1) and ATP-binding cassette super-family G member 2 (ABCG2) in CTCs [83].

CTCs can be isolated from individual patients and grafted into mice (CTC-Derived Xenografts, CDX) in order to create a patient-specific model on which specific therapies can be tested, biomarkers of response can be evaluated, and resistance mechanisms can be detected [84]. Grillet et al., based on the hypothesis that some CTCs can develop the CSC phenotype, generated multiple cell lines of CTCs from mCRC patients. They observed how the CTC lines expressed CSC markers with multilineage differentiation ability, both in vitro and in vivo, with the ability to produce liver metastasis after intrasplenic injection in xenografts [83].

4. Circulating Tumor DNA

4.1. Introduction, Methodology, and Limitations

cfDNA were first detected by Mandel and Metais in 1948 in healthy patients [85]. However, only in 1989 did Stroun first report that some ctDNA in oncological patients was derived from tumor cells [86].

cfDNA consists of single- and double-stranded DNA fragments (≤200 pb) released into bodily fluids by apoptotic or necrotic cells. While cfDNA is mostly derived from normal healthy leukocytes and stromal cells, in the context of malignancies, cfDNA can also incorporate DNA released by tumor cells, known as ctDNA. ctDNA accounts for 0.01 to 90% of the total cfDNA, depending on the tumor type, biological behavior, and cancer stage [87]. Furthermore, CRC is a solid tumor shedding the highest amount of ctDNA into the bloodstream [88,89].

In recent years, several technologies have emerged to detect ctDNA, including PCR-based approaches and next-generation sequencing (NGS). In particular, NGS allows the detection of multiple genes in terms of copy number or fusions at higher cost and lower sensitivity. On the other hand, digital PCR (dPCR) presents the highest quantitative sensitivity with restricted gene coverage.

In general, cfDNA is used to discover mutations, copy number aberrations, microsatellite alterations, differential cfDNA lengths, and methylation status [90]. At present, it is believed that ctDNA may be useful for risk classification, early recurrence detection in CRC, and as a predictive biomarker of responses to systemic treatment (for adjuvant and metastatic therapy), as we are going to analyze below.

Despite the advantages of ctDNA over a tissue-based approach, logistical and biological limitations are restricting its clinical application. From a biological point of view, the first and the most relevant limitation is tumor DNA shedding, which is known to correlate with tumor burden and with the localization of tumor metastasis, ultimately affecting sensitivity and the false negative rate [91,92,93]. Furthermore, cfDNA clearance is dependent on renal function; thus, ctDNA analyses in certain patients with renal dysfunction warrants caution [94]. Another important limitation is the CHIP (Clonal Hematopoiesis of Indeterminate Potential) mutations, aberrations derived from the DNA of hematopoietic stem cells that can also affect genes implicated in solid tumors (e.g., Tumor Protein P 53, KRAS) and hematological diseases (e.g., Tet Methylcytosine Dioxygenase 2, Janus Kinase 2), causing an increase in the false positive rate [95].

4.2. Metastatic Diagnosis and Prognosis

There is increasing evidence supporting a strong concordance between metastatic disease diagnosis and high detection sensitivity of ctDNA analysis in CRC patients due to the great amount of shedding of circulating tumor fragments (including cells, DNA, methylation markers, etc.) [96]. Variability in the detection of this analyte is known to change according to tumoral burden: liver disease is associated with higher detection rates compared with either peritoneal or lung metastases. Despite certain limitations, such as the presence of barriers (ascites, pleural effusion, etc.) or clonal heterogeneity, there is evidence confirming that it is a useful tool with high sensitivity and specificity, both in diagnostic and prognostic CRC scenarios [37,96]. The positive association between the ctDNA levels and stage IV disease has been demonstrated in different studies, suggesting that a higher ctDNA concentration correlates with disseminated disease and increased tumor burden [97,98]. Olmedillas-López et al. performed whole-exome sequencing of ctDNA in metastatic and non-metastatic patients, showing a higher level in the former group, although it was not statistically significant; in addition, a “Differential Presence of Exone (DPE)” was identified, allowing the distinction between the two groups [99]. Another study reported ctDNA levels according to tumor stage: patients with stage IV had significantly higher values than those with stage I (p = 0.0149), with the increase being directly proportional to the size of the tumor [100].

Regarding the prognosis of mCRC patients, a meta-analysis including 1076 patients from ten studies showed a favorable OS for patients with low median ctDNA levels (HR 2.39, 95% CI 2.03–2.82; p < 0.0001). Overall, cfDNA levels above the median value were significantly associated with poor prognosis (HR 2.01, 95% CI 1.54–2.62; p < 0.0001) [101]. A post hoc analysis of the NORDIC-VII study showed in mCRC patients that those with a cfDNA above the predefined normal limit had a worse mOS than those with a lower value (16.6 months vs. 25.9 months, HR 1.83, 95% CI 1.51–2.21; p < 0.001) [102]. ctDNA was also found to play a prognostic role in oligometastatic operated patients. In a study of 54 patients with Resectable Colorectal Liver Metastases (CRLMs), those with detectable ctDNA in the postoperative setting had significantly worse Recurrence-Free Survival (RFS) (HR 6.3; 95% CI 2.58–15.2; p < 0.001) and mOS (HR 4.2; 95% CI 1.5–11.8; p < 0.001) than patients with undetectable ctDNA [103]. A more recent prospective study confirmed this trend. Of 96 patients with resected liver metastases, those with postoperative ctDNA or undergoing adjuvant chemotherapy had a lower RFS than those without ctDNA (HR 4.5, p < 0.0001 and HR 8.4, p < 0.0001). Furthermore, the ctDNA status proved to be a stronger predictor of recurrence than standard clinical risk factors and CEA levels [104].

A diagnostic and prognostic role in mCRC has also been studied with the hypermethylation of genes identified by ctDNA analysis [105]. Some methylated ctDNA markers have been reported as predictors of disease prognosis of mortality, such as ranched Chain Amino Acid Transaminase 1 (BCAT1), IKAROS Family Zinc Finger 1I (KZF1), and Interferon Regulatory Factor 4 (IRF4) [106]. Picardo et al. demonstrated how downregulated expression and hypermethylation of the Beta-1,4-Galactosyltransferase 1 (B4GALT1) promoter has a negative prognostic impact on mCRC [107]. Philipp et al. detected a lower OS in mCRC patients with the methylation of helicase-like transcription factor (HLTF) or hyperplastic polyposis 1 (HPP1), or elevated levels of CEA (p < 0.0001 for all) [108].

4.3. Therapeutic Response Evaluation

Several studies have shown that early changes in ctDNA levels can predict the radiological response to standard chemotherapy in mCRC. Tie et al. demonstrated, in 53 patients treated with first-line chemotherapy, that a fold reduction of ≥10 in ctDNA at two weeks of treatment predicted the clinical response at a CT scan performed at eight weeks (p = 0.016). The decline in this analyte from the pre-treatment period to the second cycle of therapy was the best predictor of the radiological response (p = 0.004 vs. p = 0.253 of absolute ctDNA) [109]. In the prospective PLACOL trial, mCRC patients treated with a first- or second-line treatment who had a ctDNA reduction of ≥80% achieved a better Objective Response Rate (ORR) (47.1% vs. 0%; p = 0.03), better mPFS (8.5 months vs. 2.4 months; HR 0.19, 95% CI 0.09–0.40; p < 0.0001), and mOS (27.1 months vs. 11.2 months; HR 0.25, 95% CI 0.11–0.57; p < 0.001) [110]. Studies were also performed to assess the response to regorafenib and TAS-102 by ctDNA analysis. Patients who had an early decline in this analyte in the plasma experienced better PFS in contrast with those with a minimal change or increase in the ctDNA levels [111,112,113].

In addition, ctDNA liquid biopsy has been shown to play an important role in evaluating responses to immunotherapy and, in particular, in assessing pseudoprogression. This event could manifest at the beginning of immunotherapy treatment in some solid tumors and not be related to disease progression [114]. There are some prospective trials that will hopefully provide more knowledge regarding the optimal use of ctDNA in those cases. In melanoma patients treated with anti-PD-1 antibodies, ctDNA showed a sensitivity of 90% (95% CI, 68–99%) and a specificity of 100% (95% CI, 60–100%) in predicting pseudoprogression [115]. In a study evaluating treatment with regorafenib and nivolumab or pembrolizumab in 18 pre-treated Microsatellite-Stable (MSS) mCRC patients, the increase in ctDNA level at 4 weeks predicted progression at 2 months, and the decline predicted disease stability [116].

Aberrant hypermethylation also plays an important role in measuring ctDNA and the treatment response in mCRC. There are also some reports [110] describing a correlation between decreasing levels of methylated genes and response to treatment, even before radiological findings (RECIST criteria). Furthermore, hypermethylated neuropeptide Y circulating tumor DNA (meth-NPY) has recently been studied as an early indicator for a loss of treatment effect and early indicator of progression in mCRC patients in the FOLFOXIRI-Toco trial [37,117].

4.4. Detection of RAS Mutations

KRAS (exon 2, 3, or 4) and NRAS (exon 2, 3, or 4) mutations are detected in approximately 44% and 4% of patients with mCRC [4]. These are known to be negative predictive biomarkers for anti-EGFR antibody therapies, as already demonstrated in numerous prospective and retrospective trials [118,119]. Several studies have demonstrated the possibility of determining RAS mutations by ctDNA using new technologies. Taly et al. investigated the utility of multiplex dPCR to screen for the seven most-common mutations in codons 12 and 13 of KRAS from the plasma of mCRC patients. Out of 50 patients, 19 were identified to have these alterations by ctDNA analysis and 14 coincided with tissue alterations; conversely, 2 KRAS mutations were identified among those who were KRAS WT by tissue analysis [120]. In a prospective study of 106 patient samples, a quantitative PCR-based method to detect the seven point mutations of KRAS showed a specificity of 98% and a sensitivity of 92%, with a concordance value of 96% [121]. Bettegowda et al. demonstrated a ctDNA sensitivity and specificity of 87.2% and 92.2%, respectively, in determining KRAS mutations in advanced CRC patients [88]. Vidal et al. conducted an OncoBEAM analysis of RAS alterations in tissue and plasma samples from 115 mCRC patients and found 93% overall agreement; the authors identified the site of metastasis (peritoneal and lung), mucinous histology, and prior treatment as factors negatively impacting RAS detection in ctDNA [122]. Two other analyses showed a concordance of 97% and 91.8%, respectively [123,124], while Grasselli et al. identified a 90% agreement between standard of care (SoC) PCR techniques and dPCR (BEAMing) for plasma molecular analysis [125]. A prospective trial conducted on 425 chemotherapy-naive patients with liver metastases showed an accuracy in determining RAS status of 93.5% with NGS and 97% with NGS plus methylated biomarkers [126]. The high concordance rates from these studies may allow clinicians to use the analysis of the plasma RAS status to select patients who are WT and thus sensitive to anti-EGFR treatment. In the CAPRI-GOIM trial, the status of RAS mutations in ctDNA was retrospectively investigated in KRAS WT patients who received first-line treatment with FOLFIRI plus cetuximab; WT patients had better PFS and OS than those with mutations considering both tissue and liquid biopsy [127]. ctDNA is also effective in detecting primary resistance mutations to anti-EGFR antibody therapies. Siravegna et al. retrospectively analyzed the ctDNA of WT patients refractory to anti-EGFR agents with an NGS panel of 226 genes. Molecular alterations related to intrinsic resistance to cetuximab or panitumumab, such as aberrations in EGFR, Fms-like Tyrosine Kinase 3 (FLT3), Mitogen-activated Protein Kinase Kinase 1 (MAP2K1), and ERBB2, were detected in about 50% of the study population [123].

The change in the RAS mutation status in ctDNA could also play a predictive role in predicting the therapeutic response in CRC patients. In a study conducted by Sunakawa in 62 mutated RAS patients, the FOLFOXIRI + bevacizumab triplet exhibited an mPFS of 12.1 months, an mOS of 30.2 months, and an ORR of 75.8%. In 78% of patients with pre-treatment ctDNA RAS-positive disease, post-treatment ctDNA analysis at 8 weeks showed that RAS mutations disappeared, predicting better outcomes than those with RAS mutation persistence on liquid biopsy (mOS 16 months vs. 27.2 months, respectively; HR 0.53, 95% CI 0.23–1.21, p = 0.12) [128]. The KRASp.G12C mutation accounts for around 10% of KRAS mutations in mCRC, and several specific inhibitors have demonstrated efficacy in terms of response [129,130]. In a 5-year retrospective analysis conducted by Thein et al., a 3.5% rate of KRASp.G12C mutation was found in the ctDNA of mCRC patients, with a concordance of 100% and 62% to tissue molecular analysis, considering times between the two approaches of less or more than 6 months, respectively [131].

4.5. Rechallenge with Anti-EGR Antibodies

It is known that mCRC RAS and BRAF WT patients are sensitive to the EGFR blockade, but often, despite an initial response to treatment, acquired resistance mutations arise. These aberrations are mostly activating mutations of RAS and point mutations of the extracellular domain of EGFR (ECD), although amplifications of Mesenchymal-epithelial Transition Tyrosine Kinase Receptor (MET) and ERBB2 are sometimes detected [123,132,133]. It has been shown that KRAS and EGFR ECD mutations may be non-mutually exclusive [132]. In this context, ctDNA can easily and effectively identify multiple mechanisms of resistance in the same patient. Recently, Topham et al. identified a panel of 12 genes via ctDNA with an increased frequency of mutations after anti-EGFR therapy [134]. In a series of 24 patients, approximately 40% of them developed KRAS mutations after panitumumab monotherapy; the median time of onset of these aberrations was between 5 and 6 months following treatment, indicating that they were present in expanded subclones before therapy [135]. Van Burgher et al. showed that patients who developed EGFR ECD variants had a longer and greater response than those with the emergence of RAS mutations, who experienced limited responses and worse PFS. Furthermore, the detection of RAS aberrations prior to exposure to anti-EGFR antibody therapy compared with EGFR ECD mutations indicates that the latter are de novo resistance mutations [136].

Therefore, under the selective pressure of anti-EGFR antibody treatment, pre-existing resistant subclones may expand and de novo mutations may develop, leading to resistance and disease progression. In the meantime, subsequent therapy could reduce the proliferation of resistance clones, potentially restoring anti-EGFR therapy sensitivity (122). Parseghian et al. pointed out that, after the discontinuation of panitumumab or cetuximab, mutant RAS and EGFR clones decay with a half-life of 4.4 months [137]. The early identification of resistance mutations by ctDNA could allow the discontinuation of an ineffective treatment, complementing radiological assessments. In addition, numerous studies have retrospectively and prospectively evaluated the role of ctDNA-guided rechallenge with anti-EGFR antibodies in CRC patients (

Table 2. Principal rechallenge studies with anti-EGFR antibodies in mCRC.

Trial Name and Author	Study Type	Rechallenge Treatment	Line	N	OS	PFS	ORR	ctDNA Selection
Santini et al.	Retrospective	Irinotecan + cetuximab	≥3rd	39	NR	6.6 m	53.8%	None
Liu et al.	Retrospective	Cetuximab ± erlotinib	≥2nd	89	NR	4.9 m for prior responder vs. 2.5 m for prior no-responder (p = 0.064)	NR	None
Tanioka et al.	Retrospective	Irinotecan + cetuximab	≥3rd	14	NR	4.4 m	21.4%	None
Rossini et al.	Retrospective	FOLFIRI + cetuximab/FOLFOX + panitumumab/CapIri + cetuximab/Irinotecan + panitumumab/Irinotecan + cetuximab/Cetuximab/Panitumumab	≥3rd	86	10.2 m	3.8 m	19.8%	None
Karani et al.	Retrospective	Cetuximab ± CT	≥3rd	17	7.5 m	3.3 m	18%	None
Chong et al.	Retrospective	Anti-EGFR ± CT	≥2nd	22	7.7 m	4.1 m	4.5%	None
CAPRI-GOIM (Ciardello et al.)	Prospective	FOLFOX + cetuximab vs. FOLFOX (PD after FOLFIRI + cetuximab in WT pt)	2nd	66 (pt WT in retrospective NGS analysis)	23.7 m vs. 19.8 m (HR 0.57, 95% CI 0.32–1.02; p = 0.056)	6.9 m vs. 5.3 m (HR 0.56, 95% CI 0.33–0.94; p = 0.025)	29.4% vs. 9.4%	None
CRICKET (Cremolini et al.)	Prospective	Irinotecan + cetuximab	3rd	28	9.8 m (12.5 m for ctDNA WT vs. 5.2 m for ctDNA M; HR 0.58, 95% CI 0.22–1.52; p = 0.24)	3.4 m (4 m for ctDNA WT vs. 1.9 m for ctDNA M;HR 0.44, 95% CI 0.18–0.98; p = 0.03)	14% (all ctDNA WT)	Retrospective analysis
PACER (Piccirillo et al.)	Prospective	Panitumumab	≥2nd	41 (all WT before rechallenge)	6.8 m	2.1 m	7.3%	None
JACCRO CC-08 (Masuishi et al.)	Prospective	Irinotecan + cetuximab	3rd	34 (all baseline WT)	8.2 m	2.4 m	2.9%	None
JACCRO CC-09 (Tsuji et al.)	Prospective	Irinotecan + panitumumab	3rd	25 (all baseline WT)	8.9 m	3.1 m	8.3%	None
Sunakawa et al.	Retrospective(post-hoc analysis of JACCRO CC-08 and CC-09)	Irinotecan + anti-EGFR	3rd	16 (all baseline WT)	8.9 m (3.8 m for ctDNA M vs. 16 m for ctDNA WT; HR 12.4, 95% CI 2.7–87.7; p = 0.0028)	3.1 m (2.3 m for ctDNA M vs. 4.7 m for ctDNA WT; HR 6.2, 95% CI 1.6–30.5; p = 0.013)	0%	Retrospective analysis
CAVE (Martinelli et al.)	Prospective	Avelumab + cetuximab	3rd	77 (all baseline WT)	11.6 m (17.3m for ctDNA WT vs. 10.4 m for ctDNA M; HR 0.49, 95% CI 0.27–0.90; p = 0.02)	3.6 m (4.1 m for ctDNA WT vs. 3 m for ctDNA M;HR 0.42, 95% CI 0.23–0.75; p = 0.004)	7.8% (8.5% for WT vs. 5.1% for M)	Retrospective analysis
BEYOND (Aparicio et al.)	Prospective	FOLFIRI + panitumumab vs. FOLFIRI	2nd	31 (all WT ctDNA before retreatment)	13 m vs. 10 m (HR 0.55, 95% CI 0.2–1.48)	11 m vs. 4 m (HR 0.58, 95% CI 0.25–1.3)	33% vs. 7.7%	Interventional
CHRONOS (Sartore-Bianchi et al.)	Prospective	Panitumumab	≥3rd	27 (all WT ctDNA before rechallenge)	55 wks	16.4 wks	30%	Interventional
E-RECHALLANGE (Nakamura et al.)	Prospective	Irinotecan + cetuximab	≥3rd	33 (all WT at baseline)	8.6 m	2.9 m (7 m for ctDNA WT vs. 2.9 m for ctDNA M)	15.6% (50% in ctDNA WT)	Retrospective analysis
Montagut et al.	Prospective	Sym004 (futuximab + modotuximab) 12 mg/Kg (arm A) vs. Sym004 6 mg/Kg (arm B) vs. SoC (arm C)	≥3rd	254 (all WT at baseline and acquired resistance to prior anti-EGFR therapy)	7.9 m vs. 10.3 m vs. 9.6 m (HR 1.31, 95% CI 0.92–1.87 for A vs. C; HR 0.97, 95% CI 0.68–1.4 for B vs. C) In ctDNA WT: 10.6m vs. 12.8m vs. 7.3m	2.8 m vs. 2.7 m vs. 2.6 m	14.1% vs. 9.6% vs. 2.9%	Retrospective analysis
Mariani et al.	Retrospective	Irinotecan + cetuximab or cetuximab	≥3rd	26 (all WT ctDNA before rechallenge)	5 m	3.5 m	25%	Retrospective analysis
D’Onofrio et al.	Prospective	CT + anti-EGFR	≥3rd	10 (all WT ctDNA before rechallenge)	NR	11.3 m	70%	Interventional

). The CAPRI-GOIM trial showed that KRAS, NRAS, BRAF, and PIK3K WT patients, from tissue biopsy at baseline, benefited in terms of PFS from rechallenge with cetuximab [138]. Santini et al. prospectively studied the efficacy of a rechallenge with cetuximab plus irinotecan-based therapy in 39 heavily pre-treated patients. The mPFS was 6.6 months and the ORR was 53.8%, of which CR accounted for 5.1% and PR accounted for 48.7% [139]. These results were confirmed by another phase II trial where third-line rechallenge with irinotecan plus cetuximab demonstrated a PFS and OS of 2.4 and 8.2 months, respectively [140]. The role of ctDNA in guiding the rechallenge strategy was first demonstrated in the CRICKET trial. The 28 mCRC RAS and BRAF WT patients enrolled achieved 21% ORR and 54% disease control rate (DCR) with a third-line treatment of irinotecan plus cetuximab. No RAS mutations were identified with ctDNA in the patients who obtained a PR. Furthermore, WT ctDNA patients had a better PFS than those with RAS-mutated ctDNA (4 months vs. 1.9 months; HR 0.44, 95% CI 0.18–0.98; p = 0.03) [141]. The CAVE trial demonstrated the efficacy of the combination of cetuximab and avelumab as a rechallenge. RAS/BRAF WT ctDNA patients had an OS of 17.3 months compared with 10.4 months in ctDNA-mutated patients (HR 0.49, 95% CI 0.27–0.90; p = 0.02); this advantage was also confirmed in terms of PFS (4.1 months vs. 3 months; HR 0.42, 95% CI 0.23–0.75; p = 0.004) [142]. A phase II study demonstrated the efficacy of Sym004, a combination of two anti-EGFR antibodies, futuximab and modotuximab, in mCRC patients who had progressed to either panitunumab or cetuximab and maintained a triple WT status for RAS, BRAF, and EGFR ECDs [143]. The first study to investigate the prospective role of ctDNA in guiding the rechallenge strategy was the CHRONOS trial. RAS/BRAF WT mCRC patients in tumor tissue analysis, with a PS ECOG 0–2, should have obtained a PR or CR with an anti-EGFR-based first-line treatment. After progression to a second anti-EGFR-free regimen, patients underwent interventional ctDNA-based screening, and those without acquired resistance mutations received treatment with panitumumab. Of 52 patients, 16 (31%) were excluded because they had at least one alteration that conferred resistance to anti-EGFR therapy. Twenty-seven WT patients were enrolled, and 30% achieved a PR, with a DCR of 63%; a PFS and OS of 16 and 55 weeks were also recorded, respectively [144]. Data from the phase 2 BEYOND study have recently been published. Thirty-one WT mCRC RAS patients who received first-line treatment with FOLFOX plus panitumumab and who persisted WT in ctDNA after progression received second-line treatment with FOLFIRI plus panitumumab (arm A) versus FOLFIRI alone (arm B). Despite closure for inadequate recruitment, mPFS of 11 months versus 4 months in favor of the anti-EGFR arm (HR 0.58, 95% CI 0.25–1.3) and an ORR of 33% and 7.7%, respectively, were observed; however, serious adverse events were more frequent in arm A (44% vs. 23%). At progression, RAS mutations were detected by liquid biopsy in 36% and 20% of patients, respectively [145].

There are several ongoing trials investigating the efficacy of rechallenge with anti-EGFR compared with standard treatment, with some of those involving the use of ctDNA in an interventional manner. In the multicenter phase II CAPRI 2 GOIM study (NCT05312398), 200 RAS/BRAF WT mCRC patients will be treated with FOLFIRI plus cetuximab in the first-line treatment. At progression, those with acquired resistance mutations in ctDNA will receive second-line with FOLFOX plus bevacizumab; conversely, patients with persistent WT status will continue to receive anti-EGFR combined with FOLFOX. Finally, at later disease progression, the third line will consist of regorafenib or trifluridine/tipiracil for ctDNA-mutated patients and irinotecan plus cetuximab for ctDNA WT patients. (

Table 3. Ongoing trials with interventional use of ctDNA in mCRC.

Name (NCT Number)	Phase	Setting	Line	Treatment Arms	N (Actual/Estimated Enrollment)	Primary Endpoints	Status
PULSE (NCT03992456)	2	Rechallenge	3rd	Panitumumab vs. regorafenib or trifluridine/tipiracil	120	OS	Active, not recruiting
PARERE (NCT04787341)	2	Rechallenge	3rd–4th (sequence strategy)	Panitumumab followed by regorafenib vs. regorafenib followed by panitumumab	214	OS	Recruiting
CAPRI 2 GOIM (NCT05312398)	2	Rechallenge/reintroduction	2nd–3rd (1L FOLFIRI + Cetuximab)	2L FOLFOX + cetuximab (ctDNA WT) or FOLFOX + bevacizumab (ctDNA M) 3L irinotecan + cetuximab (ctDNA WT) vs. regorafenib or trifluridine/tipiracil (ctDNA M)	200	ORR	Recruiting
NCT04775862	2	Rechallenge	3rd	Anti-EGFR (ctDNA WT) or SoC (ctDNA M)	60	ORR and PFS	Recruiting
CITRIC (EudraCT Number:2020-000443-31)	2	Rechallenge	3rd	Irinotecan + cetuximab vs. regorafenib or trifluridine/tipiracil	66	ORR	Recruiting
PURSUIT (jRCTs031190096)	2	Rechallenge	3rd	Irinotecan + panitumumab	50	ORR	Not yet recruiting
NCT04509635	3	Rechallenge	3rd (non-resectable liver metastases)	Cetuximab + CT vs. CT	50	DCR	Not yet recruiting
NCT03844620	2	Response to treatment	≥3rd	Regorafenib or trifluridine/tipiracil	100	Early change in ctDNA as a predictor of radiological progression; safety	Recruiting
NCT04831528	2	Target therapy	Progression at cetuximab-based therapy	2L target therapy according to ctDNA analysis	100	ORR	Not yet recruiting
FOLICOLOR (NCT04735900)	NA	Response to treatment	1 L (WT)	FOLFOX/FOLFIRI + panitumumab	60	To evaluate response and progression by NPY methylation(ctDNA)	Recruiting
LIBImAb (NCT04776655)	3	Efficacy of treatment	1 L (WT on solid tumor biopsy but M at liquid biopsy)	Bevacizumab + FOLFIRI vs. cetuximab + FOLFIRI	280	PFS	Recruiting
COPERNIC (NCT05487248)	NA	Response to treatment	≥3rd	SoC	103	To select timepoint and cut-off value for early on-treatment ctDNA changes	Not yet recruiting
OPTIMISE (NCT04680260)	2	Treatment selection and follow up	Oligometastatic CRC treated with local therapy (escalation or de-escalation CT/observation)	ctDNA-guided treatment approach vs. SoC	350	Recurrence-free rate	Recruiting
NCT05495672	NA	Treatment selection and follow up	mCRC with metastatic small pulmonary nodules (local therapy or observation)	ctDNA-guided treatment approach	100	PFS	Recruiting
NCT03436563, cohort D	1b/2	Treatment selection	Oligometastatic CRC MSI with positive ctDNA following resection of liver metastases	Anti-PD-L1/TGFbetaRII fusion protein M7824	NA	Clearance ctDNA	Active, not recruiting
NCT04555369	NA	Response to treatment	mCRC receiving CT	ctDNA testing	300	ORR	Recruiting
NCT05141721	2/3	Response to treatment	Maintenance therapy in mCRC pt after SoC	GRT-C901/GRT-R902 (neoantigen vaccine) + ipilimumab + atezolizumab + fluoropyrimidine + bevacizumab vs. fluoropyrimidine + bevacizumab	665	Antitumor activity by number of pt with ≥50% decrease from baseline in ctDNA; PFS	Recruiting

Abbreviations: mCRC: metastatic colorectal cancer; ctDNA: circulating tumor DNA; OS: overall survival; PFS: progression-free survival; ORR: objective response rate; DCR: disease control rate; NA: not available; MSI: microsatellite instability; NPY: neuropeptide Y; PD-L1:programmed cell death ligand 1; TGF: transforming growth factor; WT: wild type; M: mutated; pt: patients; CT: chemotherapy; FOLFOX: 5-fluorouracilo + oxaliplatino; FOLFIRI: 5-fluorouracilo + irinotecan; FOLFOXIRI: 5-fluorouracilo + oxaliplatino + irinotecan; SoC: standard of care; EGFR: epidermal growth factor receptor.

).

4.6. Detection of Other Mutations (MSI, BRAF, MET, and ERBB2)

4.6.1. MSI

Microsatellites are repetitive DNA sequences ranging in length from one to six bases spread throughout the whole human genome, and they are prone to DNA replication errors. During the process of DNA replication, deletions or insertions are introduced at these sites, most of which are usually corrected by the DNA Mismatch Repair System (MMR). If the MMR is damaged, these deletions or insertions are accumulated, producing length polymorphisms of microsatellites known as MSI. Depending on their frequency, tumors are divided into three subtypes: high-MSI (>30%), low-MSI (0–30%), and microsatellite-stable (MSS) [146]. Low-MSI and MSS tumors are classified as a single type in clinics [147]. Nowadays, 12–20% of CRCs are explained by MSI-H. Its incidence is higher in the early stages (20% in stages I and II, and 12% in stage III), and a lower incidence in stage IV (about 5%) [73,148]. During the last years, the MSI has shown implications in prognosis and as a predictive biomarker of the response to immune checkpoint inhibitors [149,150,151]. Currently, MSI detection is being carried out on tumoral tissue by immunohistochemistry (IHC) and PCR, with the most widely used commercial kit being the “pentaplex assay” [152,153]. Owing to the current trend toward the systematic detection of MSI in patients with mCRC and the limitations of these techniques in tissue, non-invasive diagnostic methods are being developed using bodily fluids. Furthermore, the use of droplet-digital PCR (ddPCR) has increased, by at least two times, the detection threshold of the gold standard (pentaplex assay) [154,155,156]. In 2020, a study demonstrated that MSI detected by ddPCR in blood samples (ctDNA) from colorectal cancer patients showed a clinical specificity and accuracy of 100% [154].

On the other hand, NGS allows the examination of microsatellites at thousands of loci simultaneously, while also obtaining the mutational profile across targeted regions in a single assay [157]. Furthermore, the NGS may quantify MSI reaching a sensitivity down to 0.05%. In 2019, another study performed ctDNA testing with a Guardant360 NGS kit. In patients, the test detected 87% (71/82) of tissue MSI-H and 99.5% (863/867) of MSS for an overall accuracy of 98.4% (934/949) and a positive predictive value of 95%. Moreover, the concordance of ctDNA MSI with tissue PCR and NGS was significantly higher than that of IHC [158].

At present, one observational prospective trial is aiming to determine the concordance between the electrophoretic mobility profiles of microsatellite biomarkers in cfDNA versus primary tumor tissues in patients with MSI CRC (NCT0359448).

4.6.2. BRAF

BRAF is a protein kinase downstream of RAS in the RAS-RAF-MEK-ERK kinase pathway. Mutations in BRAF are present in approximately 10% of patients with mCRC [159].

Nowadays, in metastatic CRC assessment, the BRAF status is mandatory before applying systemic treatment due to its therapeutic implications. The BRAF V600E mutation is described as a biomarker of poor prognosis in CRC and a significant predictor of resistance to EGFR treatment [160,161]. However, non-V600E BRAF mutations occur in about 2% of all patients with mCRC and mostly define a clinically distinct subtype of CRC with good prognosis [162].

Related to the treatment of BRAF-mutant CRC tumors, the Beacon trial demonstrated that patients with the BRAF V600E mutation were highly sensitive to doublet or triplet therapy with EGFR and MAPK kinase pathway inhibitors over the standard-of-care chemotherapy [163].

In 2014, the analysis of the BRAF V600E mutation in ctDNA was validated, showing nearly 100% specificity and sensitivity for this specific mutation [121]. Conversely, a phase I/II study showed that the BRAF V600E mutant fraction burden in ctDNA was more markedly reduced by week 4 in responder patients than in non-responder patients when they received dabrafenib, trametinib, and panitumumab combination therapy [164]. Another phase IB trial by Hong et al. in patients treated with vemurafenib, cetuximab, and irinotecan confirmed the role of BRAF V600E ctDNA allele fraction reduction in predicting the radiological response to these therapies [165]. In a study of V600E-mutated mCRC patients treated with vemurafenib + cetuximab + irinotecan, ctDNA liquid biopsy identified resistance-related genetic alterations, such as the MAPK (KRAS, NRAS, and ERBB4), PI3K (PIK3CA and PIK3R2), and receptor tyrosine kinase (Platelet-Derived Growth Factor, PDGFRB) pathways; moreover, resistance mutations were also found in TGF-β pathways (TGFBR2 and SMAD4) [166].

4.6.3. MET

MET or c-MET belongs to the family of receptor tyrosine kinases (RTKs) that is encoded by the MET proto-oncogene located on human chromosome 7 (7q21-31) [167]. The cMET-hepatocyte growth factor/scatter factor (HGF/SF) pathway plays a crucial role in several biological activities, such as motility, proliferation, cell survival, embryogenesis, angiogenesis, and wound healing [168].

A meta-analysis suggested that patients diagnosed with stage III and IV illness had higher MET expression levels compared with those diagnosed with stage I–II [169]. However, in a large cohort of mCRC, the frequency of MET amplifications was detected in 1.7% (10/590) of tumor tissue biopsies, without differences between primary and metastatic lesions [170]. These results were consistent with the data generated by The Cancer Genome Atlas [171].

Despite the low prevalence of de novo MET amplification in mCRC, acquired MET amplification was detected in up to 23% of RAS WT patients who had been treated with anti-EGFR treatments and showed disease progression [170]. This detection was performed in ctDNA.

In the future, MET amplification could be used as a potential biomarker to identify acquired resistance in patients treated with anti-EGFR therapy in mCRC [170,172]. This approach may drive the early initiation of MET inhibitors in patients who respond to cetuximab and panitumumab and do not display the emergence of KRAS mutations in blood tests during anti-EGFR therapy. In a phase 1b study published in 2020, the combination of anti-EGFR and MET inhibitor showed disease control in up to 46% of patients [173].

4.6.4. ERBB2 (HER2)

ERBB2 is a part of the family of epidermal growth factor receptors (ERBB); this family represents a group of RTKs. The best-known pathogenic mechanisms involved in ERBB2 aberrant activation are overexpression and activating mutations, both described in CRC.

The reported rates of ERBB2 positivity have varied widely in different studies due to differences in antibody clone selection, scoring criteria, staining platform, and cohort composition. ERBB2 amplifications have been observed in approximately 3% of patients with mCRC [174]. Recently, a validated scoring system was developed, HERACLES. In these diagnostic criteria, the pattern of expression, intensity of staining, and percentage of positive cells are used to define positivity. Furthermore, in equivocal cases, which are defined by moderate expression in ≥50% or 3 + ERBB2 in more than 10%, but less than 50% of tumor cells, they require in situ hybridization (FISH) to define ERBB2 overexpression [174]. On the other hand, NGS can also provide information on ERBB2 and other drivers. Some authors suggest that a Copy Number Variant (CNV) of ≥5.0 found in NGS can be diagnosed as ERBB2+. However, if the CNV is between 4.0 and 4.9, the diagnosis should be confirmed by IHQ or FISH [175].

Related to ctDNA detection, in the HERACLES trial, ctDNA sequencing by the Guardant360 assay correctly identified 96.6% of samples as ERBB2-amplified. Furthermore, to improve the diagnosis in ctDNA, the authors developed an adjusted plasma copy number (apCN) in order to correct for variations in the plasma tumor fraction between samples. This apCN showed a strong correlation with the tissue ERBB2 copy number [176]. These positive correlations between tissue and ctDNA could suggest serial determinations of ctDNA to monitor the response to treatment and to elucidate resistance mechanisms.

4.7. Cell-Free DNA Fragmentomics

One of the new frontiers in liquid biopsy is cell-free DNA fragmentomics. cfDNA fragments are DNA molecules that are released into the bloodstream by the apoptosis or necrosis of cells. In 2015, Ivanov identified the non-random fragmentation of cfDNA as a possible representation of epigenetic regulation [177]. Subsequently, Cristiano et al. highlighted how different fragmentation profiles may differ between individuals with cancer and healthy individuals [178]. Different fragmentomic markers are being studied: the size profile, preferred end, jagged end, end motif, and nucleosome footprint. Jiang et al. discovered that the length of cfDNA fragments in cancer cells is shorter than that in normal cells (dominant peaks of 143 bp and 167, respectively) [179]. The preferred ends represent genomic coordinates where cfDNA cleavage preferentially occurs, being able to be useful in the detection of early stage neoplasms. End motifs constitute the sequence of nucleotides proximal to the 5’ end of DNA, correlating with nuclease-specific activity. For example, the CCCA end motif has been shown to be at low levels in hepatocarcinoma patients compared with healthy subjects [180]. This is a phenomenon that could be related to the low expression of DNAES1L3 nuclease, whose downregulation has also been identified in other neoplasms, including colorectal cancer [180]. Another fragmentation pattern is nucleosome footprints; in fact, the disposition of nucleosomes can provide information about the origin of the tissue [181].

There are several studies that have shown how these features, which reflect different fragmentation processes, can be used in early cancer detection [182,183,184].

Wang et al. evaluated the utility of cfDNA fragmentomics in predicting the pathologic response after neoadjuvant chemoradiation therapy in patients with locally advanced rectal cancer. The model based on the 5’-end motif profile plus MRI-based tumor regression grade (mrTRG) achieved the highest cross-validation AUC (0.92, 95% CI, 0.91–0.93) [185].

In the metastatic setting, Sanchez et al. compared the fragmentomic cfDNA size profile obtained by quantitative-PCR of samples from seven healthy individuals with those of seven from mCRC patients. Although both cohorts peak at 166 bp, mCRC patients had a higher proportion of fragments between 40 and 150 bp and less between 150 and 250 bp. Furthermore, these differences directly increased with increasing mutant allele frequency (MAF). All of this indicates that the cfDNA from tumor cells has a higher degree of fragmentation and nuclease activity than that from healthy individuals [186].

5. Future Perspectives and Conclusions

The different liquid biopsy analytes (exosomes, CTCs, and ctDNA) could play an important role as a diagnostic, prognostic, and predictive tool in guiding the choice of the most appropriate therapy in the field of precision oncology for mCRC patients.

We have seen the importance of using ctDNA and CTCs for predicting prognosis in oligometastatic CRC patients [75,103,104]. This application could be considered to assess the resectability of patients with a high risk of recurrence. Similarly, there is also emerging evidence on the use of ctDNA in patients with peritoneal metastases [187], with a randomized trial evaluating ctDNA as a marker of diagnosis, early intervention, and prognosis for peritoneal metastases in mCRC compared with conventional radiological techniques currently ongoing (NCT04752930).

ctDNA could be crucial to assess the response to standard chemotherapy, as it can facilitate the early discontinuation of ineffective therapies, especially considering the last lines where preserving the patient’s quality of life is crucial [111,112,113]. In the meantime, it could provide information on the possible mechanisms of resistance in particular when using anti-EGFRs to guide clinicians in the choice of rechallenge [144]. Some data are already available about the conversion of mutated RAS status after treatment with bevacizumab-based therapy, which could generate a “WT RAS window” that would allow the subsequent use of an anti-EGFR [128,188]. The future of ctDNA in mCRC patients will be likely developed in the context of detecting less frequent molecular alterations with an already real clinical implication, such as ERBB2, MSI, and KRASp.G12C. The ongoing KRYSTAL-10 trial is investigating the efficacy of the combination of adagrasib + cetuximab compared with a standard second-line treatment in KRASp.G12C-mutated mCRC patients, with mutation detection also being performed on ctDNA (NCT04793958). Other studies testing the use of ctDNA in the mCRC setting are described in Table 3. The new frontier of liquid biopsy in mCRC will be detecting predictive biomarkers of the response to both target therapy and immunotherapy. It will be fundamental in introducing its application to clinical practice where there is already evidence, such as in the case of anti-EGFR rechallenge; this could also promote a gradual standardization of sample collection and subsequent detection techniques. Finally, a new frontier in the future of liquid biopsy in mCRC patients could be the study of cfDNA fragmentomics.

In addition, it will be essential to expand liquid biopsy use in prospective clinical trials and employ it to support preclinical research in transcriptomics and proteomics to better understand the complexity of tumor heterogeneity.