Overview of predictive and prognostic biomarkers and their importance in developing a clinical pharmacology treatment plan in colorectal cancer patients

Abstract

Introduction:

Colorectal cancer (CRC) is the second most common cause of cancer-related death worldwide. Although overall survival for CRC patients has improved with earlier screening, survival continues to vary substantially across stages. Also, while the introduction of targeted therapies, including VEGF and EGFR inhibitors, has contributed to improving survival, better tools are needed to optimize patient selection and maximize therapeutic benefits. Emerging biomarkers can be used to guide pharmacologic decision-making, as well as monitor treatment response, clarify the need for adjuvant therapies, and indicate early signs of recurrence. This is a narrative review examining current and evolving use of predictive and prognostic biomarkers in colorectal cancer.

Areas covered:

Areas covered include mutations of the MAPK (KRAS, BRAF) and HER2 pathways and their impacts on treatment decisions. In addition, novel methods for assessing tumor mutations and tracking treatment responses are examined.

Expert opinion:

The standard of care pathway for staging, and treatment selection and surveillance, of CRC will expand to include novel biomarkers in the next five years. It is anticipated that these new biomarkers will assist in decision-making regarding selection of targeted therapies and, importantly, in risk stratification for treatment decisions in patients at high risk for recurrence.

1. Introduction

Colorectal cancer (CRC) is the third most common malignancy and second most deadly cancer worldwide with an estimated 1.9 million cases and 0.9 million deaths in 2020 [1]. Survival for CRC patients has improved with increased early screening and enhanced surgical options, with an overall survival of 65% at 5 years [2]. However, survival varies by stage at diagnosis. The 5-year overall survival rate for stage I disease is 92% compared to 11% in stage IV disease [3]. Approximately 35% of patients have metastatic colorectal cancer (mCRC) at initial presentation, with 20–50% of stage II and III patients eventually progressing to mCRC [4]. The large number of patients who present with disseminated disease mandate that systemic therapies remain an integral part of CRC treatment. In that context, the emergence of targeted therapies, and the associated potential for improved disease-free and overall survival, has created a clinical need to identify molecular targets in individual patient tumors and monitor their treatment responses. The large number of patients who have presumably curative surgical treatment, but later present with metastases, implies that there is a need for markers of early tumor relapse, as well as early stage patients who might benefit from adjuvant treatment. Recently, novel biomarkers have emerged as tools that can improve risk stratification during gnostic assessment, tailor treatments with molecular specificity, monitor therapeutic responses, and identify recurrent disease. This narrative review describes several prognostic and predictive biomarkers of growing importance related to targeted therapies for CRC.

2. Methods

This is a narrative review describing current and evolving prognostic and predictive biomarkers in the treatment of colorectal cancer. An initial literature review was undertaken in PubMed using the keywords “colorectal cancer” and “biomarkers”. The initial search of the literature returned 14,000 results. Current standard of care using biomarkers also was reviewed using the National Comprehensive Cancer Network (NCCN) guidelines. After this review of the NCCN guidelines, search terms were refined to include “MSI”, “RAS”, “checkpoint inhibitors”, “BRAF”, “HER2”, and “ctDNA”. Determining KRAS/NRAS and BRAF mutation status, HER2 amplifications and MSI status are recommended by NCCN guidelines as part of the evaluation of mCRC. Checkpoint inhibitors were added to capture targeted therapy in MSI tumors and ctDNA was added both as a sampling method for mutations as well a potential prognostic marker. This further narrowed resulting reports to 570 focused on MSI, 1,200 on RAS, 159 on checkpoint inhibitors, 980 on BRAF, 200 on HER2, and 200 on ctDNA. Papers from 2001 to March 2022 were sampled with an emphasis on reports published in the last 10 years. Papers with a focus on clinical, rather than preclinical, data were selected since this review focuses on biomarkers that have been integrated into the clinical care of patients.

3. Classical staging, treatment, monitoring

Stage at diagnosis is considered the most significant predictor of survival in CRC. Traditionally, the staging of solid tumors is determined from the TNM staging system. This system is comprised of tumor size or depth (T1-4), spread to regional lymph nodes (N0-3), and the presence of any distant metastasis (M0-1). In CRC, stage I disease is defined as growth through the muscularis mucosa into either the submucosa or muscularis propria with no lymph node spread. Stage II tumors are tumors invading through the muscularis propria again with no lymph node spread compared to stage III tumors which are tumors of any depth with spread to regional lymph nodes. Stage IV tumors are tumors of any size, with or without lymph node spread, with distant metastases [5]. This staging is established after surgical resection of the tumor and its associated lymph nodes combined with cross sectional imaging to establish the presence or absence of distant metastasis.

In addition to establishing prognosis, staging is also used to plan treatment. Current recommendations are that patients with stage III (node-positive) colon cancer undergo adjuvant chemotherapy but there is some controversy about the survival benefit of chemotherapy in patients with “high risk” stage II disease [6]. The intention of this adjuvant chemotherapy is to decrease the chance of recurrence in the setting of curative resection, with the presumption that stage II (node-negative) disease has not yet metastasized. Some high risk disease-specific factors in stage II disease warranting adjuvant treatment include: less than 12 lymph nodes harvested at surgery, T4 tumors, poorly differentiated tumors, presence of bowel obstruction or perforation, and lymphovascular/perineural invasion [7]. While guidelines currently recommend that stage II patients not receive adjuvant therapy, up to 23% will have a recurrence within 5 years indicating that many patients are being under-staged by current strategies and could benefit from additional therapy [8]. The addition of biomarkers that identify these early stage high-risk patients that benefit from adjuvant therapy could remarkably improve survival in CRC.

Conventionally, monitoring of treatment response and post-surgical surveillance relies on serial radiological exams [typically computed tomography (CT)] and measurements of serum carcinoembryonic antigen (CEA) levels [9]. Currently, CEA is the only serum biomarker recommended for post-surgical monitoring of CRC. Abnormally elevated CEA triggers repeat imaging to identify recurrent or distant disease. There is controversy concerning the optimal trigger threshold for CEA in smokers, and whether it improves patient survival [10]. The relative proportion of patients with oligometastatic disease eligible for curative surgery has increased in recent years due to improving surgical and therapeutic options, making it even more crucial that distant recurrent disease is identified while these patients are still surgical candidates [11]. Beyond CEA, additional validated biomarkers could help not only in monitoring treatment response and in surveillance, but also in refining targeted treatment decisions and stratifying patients to receive adjuvant therapies. In addition to novel biomarkers, new approaches to interrogating tumors for those markers is discussed.

4. Mismatch repair, microsatellite instability, immune checkpoint inhibitors

High microsatellite instability (MSI-H) is the marker of a deficient DNA mismatch repair (dMMR) system and is present in ~15% of colorectal cancers. The mismatch repair (MMR) system corrects errors that may occur during DNA replication. However, inactivation of at least one of the DNA MMR genes (MLH1, MSH2, MSH6, PMS2) through mutation or transcriptional silencing leads to an accumulation of errors in DNA [12]. Microsatellites are short tandem repeats of single nucleotide or di-, tri-, or tetra-nucleotides in DNA sequences found throughout the tumor genome and their instability is a marker of a hypermutable phenotype. This dysfunction stems from germline mutations in MMR genes or somatic hypermethylation of CpG islands surrounding the promotor region of MLH1 and other MMR genes. Germline mutations in MMR causes hereditary nonpolyposis colorectal cancer (HNPCC or Lynch Syndrome) and accounts for ~3% of all CRCs [13]. Somatic hypermethylation of CpG islands is known as the CpG island methylator phenotype (CIMP). These CIMP tumors comprise the majority of sporadic MSI CRCs [14]. CIMP tumors can be contrasted to those with chromosomal instability (CIN), which arise in the context of the traditional pathway involving APC mutation, tubular adenoma formation, and subsequent accumulation of mutations leading to cancer [15].

MSI is more common among stage II tumors compared with stage III (lymph node-positive) CRCs and relatively uncommon in stage IV (metastatic) CRCs implying dMMR tumors are less likely to metastasize. dMMR is independently associated with improved survival compared with tumors with proficient MMR (pMMR) [16]. These dMMR tumors also have lower recurrence rates compared with pMMR tumors [17]. In addition to this difference in survival and recurrence, dMMR tumors do not respond well to 5-fluorocuracil-based adjuvant chemotherapy [18]. One explanation for this improved baseline clinical course may be that the hypermutable phenotype contributes to the production of abnormal peptides that serve as neoantigens, producing specific antitumor immune responses decreasing tumor growth and spread [19]. The distinct phenotype of CRCs with sporadic MSI include right colon predominance, increased prevalence in women and poor differentiation/mucinous histology. MSI-H tumors also exhibit an elevated number of tumor-infiltrating lymphocytes (TILs), supporting the neoantigen hypothesis [20]. While this local immune control predisposes patients with MSI-H CRC to improved survival at baseline, this phenotype also sensitizes tumors to immune checkpoint inhibitors targeting programmed cell death-1 protein (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) (Figure 1). PD-1 is expressed on T cells and binding of its ligands (PD-L1 and PD-L2) downregulates T cell effector function. Indeed, tumors escape immune detection by upregulating expression of programmed death ligand 1 (PD-L1) [21]. Inhibitors of PD-1 block this receptor from interacting with its ligands, promoting tumor cell killing by effector T cells. Inhibitors of PD-1, pembrolizumab (Keytruda) and nivolumab (Opdivo), are FDA-approved for patients with mCRC with dMMR or MSI-H and confer a significant survival benefit [22,23]. An additional immune checkpoint inhibitor target, CTL-4, is transiently expressed on activated T cells and its expression inhibits the production of cytokines and provides a negative feedback signal to T cells inducing T cell cycle arrest. Blocking CTLA-4 may lead to reactivation of T cells allowing them to overcome tumor-induced immune tolerance [24]. Ipilimumab (Yervoy) is an anti-CTLA-4 antibody that can be used in patients with metastatic dMMR/MSI-H disease in combination with nivolumab [25]. The combination of nivolumab and low-dose ipilimumab produced an objective response rate of 64%, complete response rate of 9%, and disease control rate of 84%, all of which were durable [26]. While the results of immune checkpoint blockade in dMMR/MSI-H CRC patients have been encouraging, single agent checkpoint inhibitors are not efficacious in patients with pMMR [27]. In addition, while immune checkpoint inhibitors are approved in mCRC dMMR/MSI-H disease, their utility as neoadjuvant therapy in localized disease and their efficacy in combination are being explored [28,29]. In that context, MSI status serves as both a predictive and prognostic marker.

Figure 1.

Checkpoint inhibitors. Checkpoint inhibitors such as those targeting programmed cell death-1 protein (PD-1), it’s ligand (PD-L1) or cytotoxic T-lymphocyte- associated protein 4 (CTLA-4) lead to reactivation of T cell functions allowing them to overcome tumor-induced immune tolerance. dMMR tumors have increased immune infiltration predisposing them to treatment with checkpoint inhibitors.

Neurotrophic tyrosine receptor kinase (NTRK) gene fusions are extremely rare in CRC occurring in less than 1% of tumors [30]. However, they are more frequently found in patients with dMMR [31]. The FDA has approved two tropomyosin receptor kinase (TRK) inhibitors, entrectinib and larotrectinib, for use in patients with NTRK fusion-bearing cancers in either a worsening metastatic setting or locally advanced unresectable tumors [32]. This is an example of tissue agnostic treatments that can be used based on displayed mutations rather than tissue types.

5. MAPK pathway (Ras-Raf-MEK-ERK)

Mutations in proteins in the mitogen-activated protein kinase (MAPK) pathway drive tumorigenesis. Mutations in these modulating proteins produces loss, or gain, of function leading to dysregulated proliferative cell signaling and cancer. RAS is the first protein to be activated in the pathway and commonly mutated in CRC [33]. Activation of RAS in a normal cell begins with an extracellular ligand binding to a receptor-linked tyrosine kinase like epidermal growth factor receptor (EGFR). This binding activates the tyrosine kinase in the cytoplasmic domain of the receptor leading to phosphorylation of EGFR and subsequent interaction with RAS. This binding triggers RAS, a GTPase, to exchange a GDP molecule for GTP, activating the pathway. This initiates a kinase cascade leading to the activation of Raf, MAPK/ERK (MEK1 or 2) and finally MAPK [34].

RAS (KRAS, NRAS, HRAS) is the most frequently mutated gene family in cancers and the most common oncogenic mutant of the RAS family is KRAS G12C. KRAS mutations are present in 30–50% of CRC, NRAS in 3–5%, with HRAS mutated in less than 1% [35][36]. Among KRAS mutated CRCs specifically, the breakdown among mutations is G12D 29.9%, G12V 20%, G13D 15.8%, G12A 4.9%, and Q61H 4.2% [37]. Until recently, RAS was considered “undruggable” given its picomolar affinity for GTP/GDP, the absence of known allosteric regulatory sites, and the importance of wild type RAS for normal biologic functions. However, small molecules that specifically inhibit the G12C variant have been identified, making RAS a potential therapeutic target [38].

Monoclonal antibodies targeting EGFR, including cetuximab and panitumumab, are routinely used in mCRC. These monoclonal antibodies compete with endogenous EGFR ligand and upon binding, block phosphorylation, ultimately leading to internalization and degradation of the receptor. Cetuximab has been approved as a first-line treatment in mCRC with wild-type KRAS in combination with chemotherapy [39]. There is emerging evidence of the utility of combining EGFR and KRAS G12C inhibitors in advanced disease. EGFR signaling has been identified as the dominant mechanisms of CRC resistance to KRAS G12C inhibitors [40]. The combination of anti-EGFR and KRAS G12C inhibitors is effective against cell lines, patient-derived organoids, and xenografts [41]. Unfortunately, FOLFOX, with or without cetuximab, failed to improve disease-free survival in post-resection stage III colon cancer patients [42].

In addition to being a druggable target, RAS also is a prognostic biomarker. Indeed, RAS is prognostic for worse overall survival and higher rates of lung recurrence after hepatectomy in patients with colorectal liver metastases [43]. In addition, it predicts worse radiologic and pathologic response to neoadjuvant chemotherapy [44]. Patients with KRAS mutations and lung metastases had improved overall survival and time to pulmonary recurrence when they underwent an anatomic resection instead of a wedge resection [45]. As wedge resections, rather than anatomic resections, are typically used in metastasectomy, KRAS may be useful as a predictor in determining appropriate surgical approaches in these patients.

One downstream effector target of RAS is the RAF family, comprising c-RAF1, BRAF, and ARAF. Of these, BRAF is the most frequently mutated in tumors [46]. Indeed, beyond constitutive activation of RAS, mutations in codon 600 of the BRAF gene produce RAS-independent activation of the MAPK pathway leading to increased cell proliferation and survival. Sporadic MSI CRCs often show increased co-occurrence of BRAFV600E mutations compared to CRCs overall [47]. These somatic BRAFV600E mutations increase BRAF/MEK/ERK signaling leading to the CIMP which silences MLH1, ultimately resulting in dMMR. The presence of a BRAF mutation indicates a sporadic MSI tumor and virtually excludes the diagnosis of Lynch syndrome [48]. Patients with BRAF mutations experience poorer survival compared to patients with wild-type BRAF [49].

BRAF and KRAS mutations are not coincident in tumors, and many KRAS wild type mCRC have BRAF mutations. In turn, these mutations identify tumors that are unresponsive to anti-EGFR therapies when combined with chemotherapy [50]. BRAF inhibitors are used extensively in BRAFV600E melanomas [51]. While BRAF inhibitor monotherapy in BRAFV600 melanoma leads to response rates of >50%, only ~5% of BRAFV600 CRC patients respond [52]. Since EGFR mediates resistance to BRAF inhibitors, the differential expression of EGFR in CRC, compared to melanoma, may explain differences in response rates. In CRC, BRAF inhibition leads to feedback activation of EGFR which drives proliferation even in the presence of BRAFV600 inhibition [53]. In contrast, dual blockade of EGFR and BRAF produced synergistic inhibition of tumor growth in murine CRC models through enhanced MAPK suppression [54]. Dual treatment with EGFR and BRAF inhibitors in previously-treated patients with BRAF V600E mCRC improved overall survival and progression-free survival compared to standard chemotherapy [55]. Moreover, triple therapy inhibiting BRAF, EGFR, and MEK is effective against BRAFV600 tumors [56,57]. These considerations highlight the utility of BRAF mutations as both predictive and prognostic markers.

6. HER2

HER2 (human epidermal growth factor receptor 2) is a transmembrane receptor that functions as an intracellular tyrosine kinase. Homo- or heterodimerization of HER2 with another member of the EGFR family (EGFR/HER2/ERBB) results in activation of the RAS-RAF-ERK or PI3K-PTEN-AKT pathways leading to increased proliferation (Figure 2). In quadruple WT populations (KRAS, NRAS, BRAF, PIK3CA WT) treated with anti-EGFR therapies, the HER2 pathway may function as a bypass leading to resistance to anti-EGFR agents [58]. Indeed, HER2 expression predicts unresponsiveness to EGFR-targeted therapies in mCRC [59].

Figure 2.

Epidermal growth factor signaling pathways. Multiple potential targets for therapy exist along epidermal growth factor receptor (EFGR) and human epidermal growth factor receptor 2 (HER2) pathways. In addition, amplification of HER2 has been implicated in anti-EGFR therapy resistance as activation of the HER2 pathway may bypass blockade of EGFR.

The amplification of the HER2 gene or overexpression of the protein has been targeted in other solid tumor malignancies. While therapies that block HER2 (trastuzumab, lapatinib, pertuzumab) have gained importance in treating patients with HER2-overexpressing tumors in other malignancies, there are no HER-2-directed therapies approved by the FDA to treat mCRC. Tumors with wildtype RAS and BRAF have a HER2 amplification rate of 14% [60]. In that context, treatment with HER2 inhibitors could both directly target tumor HER2 overexpression and limit resistance to EGFR inhibitors.

Preclinical work initially revealed that HER2-amplified tumors were responsive to dual HER2-directed therapies, but not individual agents alone. Using this information, a phase 2 trial examining dual HER2 therapy comprising a tyrosine kinase inhibitor and anti-HER2 monoclonal antibody in KRAS wild-type, HER2-positive mCRC patients demonstrated that 30% of patients had objective responses and 44% had stable disease [61]. A phase 2 trial of trastuzumab deruxtecan, a HER2-targeted antibody-drug conjugate, in patients who progressed on at least two previous treatment regimens, showed an objective response rate of 45.3% [62]. Given the potential to target HER2 directly or increase susceptibility to anti-EGFR agents, HER2 inclusion in mutation testing appears appropriate. In that context, HER2 serves as a predictive marker in the treatment of HER2 positive CRC by anti-EGFR agents.

7. Consensus molecular subtypes

An additional method of categorizing CRCs that may help guide selection of potential targeted pharmacologic treatments is Consensus Molecule Subtypes (CMS) (Figure 3). CMS1 or MSI immune tumors make up 14% of CRCs. These tumors have a high rate of mutations with frequent BRAF mutations and large immune infiltration. Most MSI tumors fall in this category and as discussed earlier, these tumors are responsive to immune checkpoint inhibitors. In addition, the BRAFV600E mutation confers resistance to treatment with anti-EGFR agents. CMS2 or canonical tumors constitute 37% of tumors and have upregulated Wnt/Myc signaling. These tumors arise from the canonical adenoma-to-carcinoma sequence characterized by the initial loss of APC, subsequent activating mutation in KRAS, and ultimate loss of TP53. CMS3 or metabolic tumors account for 13% of cases and have frequent KRAS mutations and dysregulation of cancer metabolic pathways. As discussed previously, KRAS mutation may indicate a poor response to anti-EGFR therapy. CMS4 or mesenchymal tumors comprise 23% of cases and are characterized by transforming growth factor beta (TGFβ) pathway activation, enhanced angiogenesis, stromal activation and inflammatory infiltrates [63].

Figure 3.

The consensus molecular subtypes. CMS1 or MSI immune have a high rate of mutations with frequent BRAF mutations and large immune infiltration. Most MSI tumors fall in this category. CMS2 or canonical tumors have upregulated Wnt/Myc signaling. CMS3 or metabolic tumors have frequent KRAS mutations and dysregulation of cancer metabolic pathways. CMS4 or mesenchymal tumors are characterized by transforming growth factor beta (TGFB) pathway activation, enhanced angiogenesis, stromal activation and inflammatory infiltrate.

These four subtypes can be broadly divided into “hot” and “cold” CRCs based on immune infiltration. The high immune infiltration of CMS1/MSI-H tumors has been discussed, as well as their predisposition to treatment with immune checkpoint inhibitors. While CMS4 tumors also have increased immune cell infiltrates, responses to immunotherapy may be affected by TGFβ signaling. In contrast to the anti-tumor immune environment of CMS1 tumors, the CMS4 tumor microenvironment is proinflammatory [64]. Indeed, TGFβ may be immunosuppressive and drive immune evasion in CRC [65]. Conversely, CMS2 and CMS3 tumors are “cold” tumors reflecting low immune cell infiltrates. CMS2 and CMS3 tumors may respond to immunogenic stimuli, like vaccines or co-stimulatory compounds, but do not respond to immune checkpoint inhibitors. CMS2 and 3 tumors also respond to anti-VEGF agents [66]. CMS classification has the potential to provide prognostic information, since CMS2 and CMS3 tumors have a better prognosis than CMS1 and CMS4. Further, CMS classification also predicts responses to immune checkpoint inhibitors given their differing immune activities. However, these classifications are not currently recommended for use in clinical practice.

8. Circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA)

The concept of a “liquid biopsy” for solid tumors has recently emerged, reflecting sampling convenience and its ability to capture the molecular makeup of a solid tumor. Liquid biopsies have several advantages compared to tissue biopsies, including capturing the heterogeneity of a tumor, assessing molecularly divergent metastatic lesions, and evaluating potential resistance mutations as treatment progresses. Circulating tumor cells (CTCs) are cancer cells that have been passively shed or intravasated from the primary lesion or metastases into the bloodstream. CTCs can be isolated from the blood of patients as either single cells or clusters [67]. The number of CTCs detected has been associated with treatment outcomes and overall survival [68]. Cell-free circulating tumor DNA (ctDNA) consists of somatic and epigenetic DNA alterations from tumors cells that are released into the bloodstream following apoptosis or necrosis. ctDNA is more abundant within the bloodstream than CTCs but both can be interrogated for actionable treatment targets and emergent resistant subclones, therefore assisting in treatment decisions before and after initiation of therapy.

Many studies have shown a link between ctDNA and tumor burden with ctDNA positivity increasing with CRC stage [69]. In addition, as discussed earlier, there are multiple molecular markers that can predict prognosis or treatment response in CRC such as MSI-H/dMMR (susceptibility to immune checkpoint inhibitors) as well as KRAS/BRAF (anti-EGFR resistance). A study examining the emergence of mutated KRAS alleles in ctDNA during anti-EGFR therapy revealed that these alleles decline when treatment is suspended, demonstrating that liquid biopsies can be used to track treatment resistance [70]. The ability to accurately capture these markers prior to treatment could help tailor therapeutic planning. In addition, the ability to track these markers during treatment could 1) ensure treatment response and 2) monitor for emerging resistance. The measurement of these biomarkers from solid tumors may be misleading given the presence of different subclones within a primary tumor or distinct metastatic lesions.

Surveillance after curative surgery in CRC consists of serial CT imaging and CEA levels to monitor for disease recurrence [71]. Also, imaging is used to monitor treatment responses in patients who receive chemoradiation either as definitive treatment or as a bridge to possible surgery. The RECIST (Response Evaluation Criteria in Solid Tumors) guidelines were developed to assist in the assessment of tumor burden in clinical trials and standardize treatment responses [72]. However, these criteria have poor reproducibility between users and is not applicable in the post-surgical patient without measurable lesions, such as mCRC patients with oligometastatic disease who undergo resection [73]. In addition, the correlation between tumor response and treatment efficacy is questionable, especially with the introduction of targeted agents [74]. RECIST guidelines consider changes in the size and number of tumors, with progressive disease (PD) defined as growth of existing lesions or the appearance of additional lesions. Traditionally, this would be considered treatment failure and the current treatment would be stopped [75]. However, these guidelines are based on specific patterns of response to cytotoxic therapies, and they may not capture the effectiveness of newer immunotherapies. Indeed, patients may initially experience what appears to be PD and treatment failure, but continued treatment may produce responses [76]. While radiologic criteria are being amended to include unique aspects of immunotherapy, the ability to monitor treatment response using ctDNA or CTCs could serve as a useful adjunct.

Currently, there is controversy as to whether stage II colon cancer patients should receive adjuvant therapy. While some high risk characteristics have been suggested (less than 12 lymph nodes harvested at surgery, T4 tumors, poorly differentiated tumors, presence of bowel obstruction or perforation, and the presence of lymphovascular/perineural invasion), there is not consensus on which stage II patients should receive adjuvant chemoradiation [77]. In stage II colon and rectal cancer, post-operative patients who were positive for ctDNA were at extremely high risk for recurrence when not treated with adjuvant chemotherapy [78,79]. A study examining ctDNA status in patients after curative-intent surgery revealed that 100% of patients with ctDNA detected after treatment completion ultimately recurred [80]. In patients with resectable colorectal liver metastases, patients with ctDNA detected after surgery had a significantly poorer relapse-free survival and overall survival. In addition, all patients with persistently detectable ctDNA after adjuvant chemotherapy recurred [81]. A study in stage I-III patients revealed that in the majority, ctDNA identified relapse after definitive treatment. The same study also showed that ctDNA status was independently associated with relapse after adjusting for other clinicopathologic risk factors [82]. ctDNA could potentially be used an adjunct to the traditional TNM staging and other potential prognostic markers in determining which patients receive adjuvant therapy, serving as both a predictive and prognostic biomarker.

9. Conclusion

While improvements in screening have enhanced disease rates and survival, CRC continues to present in younger populations, and is found only after the tumor has metastasized in more than 60% of patients. Consequently, there is a substantial vulnerable population who requires additional adjuncts to surgery. In recent years, therapies such as specific molecularly-targeted treatments and immunotherapies have become options in mCRC. While these emerging therapies are exciting, a paradigm shift will be required in how patients are selected for these therapies, and how their responses are monitored. Emerging prognostic biomarkers for risk stratification and predictive biomarkers to define treatment sensitivity or resistance will need to be integrated into the standard of care to continue improving CRC survival (Table 1).

Table 1.

Prognostic and predictive biomarkers in colorectal cancer.

Biomarker	Targeted Drugs	Resistance	Stage I-III Disease	Metastatic Disease
Microsatellite Instability	Pembrolizumab Nivolumab (PD-1 inhibitors) Ipilimumab (CTLA4 inhibitor)		dMMR tumors have a decreased chance of metastasis In stage II disease, prognostic marker of more favorable outcome Favorable impact of dMMR more limited in stage III colon cancer Stage II dMMR patients have not been shown to benefit from 5-FU adjuvant therapy	Pembrolizumab and nivolumab either as monotherapy or in combination with ipilimumab as neoadjuvant therapy
NTRK	Enterctinib Larotrectinib		Approved in locally advanced, unresectable CRC with NTRK gene fusion	Approved in mCRC with NTRK gene fusion
RAS	Small molecules targeting G12C variant	Confers anti-EGFR agent resistance	No role for anti-EGFR agents in these stages	Patients with known KRAS/NRAS mutations should not be treated with cetuximab or panitumumab (anti-EGFR)
BRAF	BRAF inhibitors	Negative predictor of response to anti-EGFR therapies	BRAF V600E mutation confers poor prognosis regardless of treatment	BRAF V600E mutation confers poor prognosis regardless of treatment
HER2	Trastuzumab Lapatinib Pertuzumab Trastuzumab deruxtecan	Predict resistance to anti-EGFR therapies	Testing recommended only in mCRC	3 regimens: -Trastuzumab derutecan monotherapy -Trastuzumab in comvination with either lapatinib or pertuzumab

10. Expert opinion

Despite improvements in screening and surgical interventions, CRC has remained the second most common cause of cancer-related death worldwide. While it has an overall favorable relative survival 5 year survival of 65%, inadequacies in treatments are revealed when stage-specific prognosis is examined [2]. The 5-year overall survival for stage I disease is 92% compared to 11% in stage IV [3]. Approximately 35% of patients have metastatic disease at presentation with 20–50% of stage II and stage III patients progressing to metastatic disease [4]. These survival statistics illuminate remarkable opportunities for improvement in the treatment of CRC. Further, the substantial fraction of patients who present with distant metastases highlight the continued unmet need for improved systemic therapies to treat mCRC. In that context, emerging molecularly targeted therapies offer additional approaches to treat disseminated disease beyond standard chemoradiation. Additionally, significant recurrence rates in patients presenting with localized disease underscore opportunities for curative treatment in some patient populations. Moreover, while there is consensus about the benefit of adjuvant therapy in patients with stage III disease (positive lymph nodes), controversy remains about those with stage II disease. Again, recurrence rates suggest that some of these patients could benefit from molecularly-targeted adjuvant therapies. These considerations bring into specific relief the clinical unmet need for the combination of improved diagnostics and therapeutics in CRC.

Emerging advances in technology have provided remarkable tools that have yielded unprecedented insights into the molecular underpinnings of CRC. These approaches have revealed substantial heterogeneity in mechanisms underlying tumorigenesis, and defined the precise molecular components and their interactions that produce different molecular subtypes of disease. In turn, identification of these mechanisms has provided targeted opportunities to better define prognostic biomarkers that define risk. In that context, KRAS-mutant tumors are more likely to recur than KRAS wild-type, regardless of stage or treatment. Similarly, patients with ctDNA or circulating tumor cells have a very high risk of developing metastatic disease. Conversely, patients with MSI-H tumors have a more favorable prognosis, reflecting their characteristic generation of neoantigens which produce specific anti-tumor immune responses.

Beyond prognosis, these mechanistic insights have provided opportunities to molecularly tailor therapies to the individual patient and tumors. However, the heterogeneity of these mechanisms underscore the need for predictive biomarkers that define the dysfunctional pathways underlying tumorigenesis, to get the right treatment for the right tumor in the right patient. Thus, checkpoint inhibitors are effective in patients with dMMR, put not pMMR, tumors, reflecting their immunological activity. Her-2 blocking antibodies are effective in Her-2 over-expressing tumors. BRAF inhibitors are effective in combination with monoclonal antibodies that block EGFR in patients with tumors driven by mutant BRAF.

In addition to prognosis and prediction, biomarkers are becoming essential for monitoring therapeutic responses to molecularly-targeted therapies. Indeed, ctDNA can be used to follow tumor regression, escape from targeted therapy, and recurrence and progression. Moreover, this molecular approach provides an opportunity to sample tumor DNA in real time during therapy to define mechanisms of therapeutic escape and resistance, and the emergence of novel mutations driving tumor progression requiring a shift in therapeutic strategy. In that context, ctDNA is emerging as a paradigm-shifting technology that is positioned to revolutionize prognostic and predictive management of CRC patients.

Over the next 5 years, emerging evolution in technology will invariably drive an acceleration of progress in identifying novel molecular mechanisms underlying the different subtypes of CRC. In turn, these insights will provide the framework to develop new therapies that target those specific molecular mechanisms. As those therapies emerge, they will drive the development of companion biomarker technologies that can identify patients that will be responsive to therapy, monitor their disease regression in response to treatment, and identify emergent therapeutic resistance. Beyond the clinical imperatives driving this co-evolution of treatments and biomarkers, there is an associated economic mandate to tailor expensive molecular therapies only to those who can best benefit from them.

Article highlights:

• Current recommendations suggest that most stage II CRC patients undergo surveillance for recurrent disease, rather than receive adjuvant therapy. Given the percentage of these patients who later develop with metastases, there is a clear unmet clinical need for additional markers of high-risk disease.

• Metastatic colon cancer with microsatellite instability respond to treatment with immune checkpoint inhibitors (PD-1 inhibitors and CTL4 inhibitors). Additional work is being conducted in patients with earlier stage disease.

• BRAF mutations are a negative predictor of susceptibility to anti-EGFR therapy. While BRAF inhibitors have had limited efficacy in CRC, dual blockade with anti-EGFR agents and BRAF inhibitors have shown success.

• The Consensus Molecular Subtype classification of CRCs may help guide therapeutic decision-making by improving prognostic and predictive categorization of tumors.

• Beyond novel mechanistic biomarkers that guide treatment decisions, innovative methods of profiling CRC tumors like circulating tumor (ct)DNA have the potential to identify patients who require adjuvant therapy and assess disease responses across the treatment continuum.