**5. CtDNA in metastatic CRC (mCRC)**

For the majority of patients with mCRC, the mainstay of treatment consists of palliative-intent systemic therapy with median overall survival (OS) approaching 24 months. Three classes of cytotoxic agents (fluoropyrimidine, irinotecan and oxaliplatin) and two classes of molecularly-targeted agents (monoclonal antibodies targeting vascular endothelial growth factor [VEGF], and the epidermal growth factor receptor [EGFR]) are currently approved for use in mCRC, although the optimal sequencing and scheduling of these treatments are debated. To optimise their therapeutic ratio and minimise toxicity, effective and accurate means of assessing treatment response are needed. In the following section, we summarise the evidence on the use of ctDNA in mCRC management. These include prognostication, monitoring tumour burden and predicting treatment efficacy, guiding targeted treatment selection, and detecting anti-EGFR therapy resistance.

#### **5.1 CtDNA in mCRC: prognostic value and monitoring tumour burden**

The association between the presence or high levels of ctDNA and adverse survival outcomes in mCRC has been demonstrated in several studies [66–69]. For example, in a landmark study, Bettegowda et al. observed a steady decrease in 2-year survival rate as ctDNA concentration increased [69]. Moreover, a systematic review exploring the prognostic role of ctDNA in CRC (mostly mCRC) found that most studies, although not all, demonstrated a negative correlation between ctDNA and disease-free survival and OS [70].

Another application for ctDNA that has been explored in mCRC is correlating longitudinal dynamics during systemic therapy with prediction of treatment response and tumour burden [71]. Currently, anatomical radiographic imaging—particularly computed tomography (CT)—is the chief modality to evaluate therapeutic benefit in mCRC. However, limitations include cost, operator- and reader-dependence, challenges in standardisation and radiographic lag behind clinical changes. Furthermore, changes in tumour size which form the basis of response measurement on CT does not account for changes in tumour density or morphology that may result from response to molecularly targeted agents commonly used in mCRC. CEA is also used in mCRC disease monitoring, usually in-between or in addition to radiology assessments. However, CEA is elevated in only approximately 70–80% of patients with mCRC and has limited sensitivity and specificity in detecting disease progression or treatment response [72].

A study of 53 mCRC patients undergoing standard first-line chemotherapy, found that significant decline in ctDNA levels using Safe-SeqS prior to cycle two chemotherapy was associated with objective radiological response at 8–10 weeks (p = 0.016) [73]. This study also found a trend between ctDNA reduction and improved progression-free survival. The more recent PLACOL study in 82 patients receiving chemotherapy for mCRC echoed these findings [7]. PLACOL utilised picodroplet-digital PCR assays based on either genomic or hypermethylation alterations. The investigators found that the baseline ctDNA concentration was prognostic for OS, and that early and deep ctDNA reductions were associated with improved objective response rate and longer survival (p < 0.001) [7]. Another recent study using digital PCR found methylation changes over time correlated with tumoural response in patients with mCRC [74].

These studies suggest that early changes in ctDNA during systemic therapy may be predictive for treatment efficacy and prognostic for survival outcomes, thus suggesting a role for serial ctDNA monitoring during palliative treatment with systemic therapy. Indeed, with the advantages of a short half-life reflecting immediate-term changes [18] and high tumour-related specificity, ctDNA monitoring may be complementary to radiological assessments and blood biomarkers currently in use. In clinical circumstances where radiological assessments are indeterminate or ambiguous, such as the lack of measurable disease by imaging criteria or the presence of mixed response, ctDNA dynamics may be of particular value; although ctDNA may not always correspond to imaging findings [75]. Equally, it is also prudent to acknowledge that no current evidence supports the strategy of biomarkermonitoring of palliative therapy and that earlier adaptive treatments will augment survival or quality of life.

## **5.2 CtDNA in mCRC: genotyping to guide targeted treatment selection**

It is now standard of care for patients with mCRC to undergo molecular profiling on their tumour tissue in order to determine *BRAF V600E* and *Rat sarcoma* (*RAS*, particularly exon 2–4 *KRAS*) gene mutational status. This informs clinical decision-making regarding benefit from anti-EGFR therapy. The advantageous role of genotyping with ctDNA has already been established in the field of advanced non-small cell lung cancer. Circulating genetic aberrations of EGFR (for example, exon 19 deletions or exon 21 (L858R) substitution mutations) are now being used in standard practice, as a companion tool, to identify eligible patients for treatment with erlotinib. This technology was approved by the US Food and Drug Administration in 2016 for this indication [76].

In mCRC, a meta-analysis of 21 studies on the diagnostic performance of ctDNA-based *KRAS* gene detection found overall high pooled specificity (96%)

**63**

*Current Utility and Future Applications of ctDNA in Colorectal Cancer*

and moderate sensitivity (67%) [77]. Not surprisingly, heterogeneity was noted to be high probably owing to varying molecular techniques, tumour stage and study designs [77]. Although a number of *KRAS* ctDNA assays have demonstrated high agreement (91–93%) with tumour tissue *KRAS* testing and are available for commercial use [78, 79]; at present, tumour tissue testing remains the gold standard to establish *KRAS* mutational status. Given the appreciable discordance rate with tumour tissue genotyping, it is recommended that a negative ctDNA result should trigger tumour tissue variant analysis [5]. As will be discussed under Section 6, ctDNA assays in mCRC may also been utilised to select predictive immune-related

**5.3 CtDNA in mCRC: genotyping to monitor for targeted treatment response** 

The role of genomic alterations and their evolution in both the development and progression of CRC have culminated in the realisation that serial genotyping of the primary tumour, and its secondaries, is ideally required if we want to succeed in personalising patient care with precision [80]. Unfortunately, patients with mCRC who do not harbour a somatic *RAS* mutation pre-treatment, will typically develop acquired resistance to anti-EGFR therapy in a matter of months after initially showing response. There is a battery of pre-clinical and clinical evidence which points to the acquisition of molecular mechanisms of resistance associated with aberrations in the *RAS*-*MEK*-*mitogen-activated protein kinases* (*MAPK*), pathway [2, 42, 81, 82]. Longitudinal ctDNA analysis can be used in this setting with high diagnostic precision to detect both primary resistance and early molecular changes that may confer acquired resistance. Several translational studies have successfully employed ctDNA to illustrate and verify the emergence of *RAS* mutations as a mechanism of

In a phase II trial of anti-EGFR antibody, panitumumab in mCRC, serial prospective plasma analysis detected more emergent *RAS* mutations than serial tissue biopsies, suggesting that the former may be more comprehensive in evaluating global tumoural heterogeneity [83]. In a small retrospective study of 10 mCRC patients who developed resistance to anti-EGFR therapy (cetuximab or panitumumab) in combination with chemotherapy, Misale et al. demonstrated that the onset of the emerging *KRAS* mutations was detected in serum ctDNA analysis as early as 10 months prior to radiological reporting of disease progression [2]. In this study, ctDNA analysis was also explored in a separate cohort of patients who were receiving chemotherapy alone (control group). No acquired *KRAS* mutations were identifiable at disease progression [2]. In the same year, Diaz et al. also demonstrated the feasibility of using serum ctDNA to identify emerging resistance to panitumumab in a prospective cohort of 28 patients [81]. Thirty-eight percent of patients whose tumours were initially *KRAS* wild type developed detectable *KRAS* mutations in their sera, three of whom developed multiple different *KRAS* mutations. These were detected approximately 5 months before radiological progression [81]. Another study subsequently made the intriguing observation that *KRAS* clones can fluctuate under the selective pressure of anti-EGFR therapy implying that there may be a role for 'pulsing' or re-challenging with anti-EGFR therapy [84]. Furthermore, in a prospective study of 108 patients, treated in the third line setting with cetuximab and irinotecan, Spindler et al. investigated the quantitative correlation between plasma cfDNA with tumour-specific plasma mutant *KRAS* levels [85]. This study revealed that (i) the majority of *KRAS* mutations that were detected in tumours were also found in the plasma, (ii) there was a strong correlation between cfDNA and plasma mutant *KRAS* levels, and (iii) high levels of plasma

*DOI: http://dx.doi.org/10.5772/intechopen.82316*

biomarkers for immunotherapy selection.

acquired resistance to anti-EGFR therapy.

**and resistance**

### *Current Utility and Future Applications of ctDNA in Colorectal Cancer DOI: http://dx.doi.org/10.5772/intechopen.82316*

*Advances in the Molecular Understanding of Colorectal Cancer*

tumoural response in patients with mCRC [74].

Administration in 2016 for this indication [76].

survival or quality of life.

response [72].

Another application for ctDNA that has been explored in mCRC is correlating longitudinal dynamics during systemic therapy with prediction of treatment response and tumour burden [71]. Currently, anatomical radiographic imaging—particularly computed tomography (CT)—is the chief modality to evaluate therapeutic benefit in mCRC. However, limitations include cost, operator- and reader-dependence, challenges in standardisation and radiographic lag behind clinical changes. Furthermore, changes in tumour size which form the basis of response measurement on CT does not account for changes in tumour density or morphology that may result from response to molecularly targeted agents commonly used in mCRC. CEA is also used in mCRC disease monitoring, usually in-between or in addition to radiology assessments. However, CEA is elevated in only approximately 70–80% of patients with mCRC and has limited sensitivity and specificity in detecting disease progression or treatment

A study of 53 mCRC patients undergoing standard first-line chemotherapy, found that significant decline in ctDNA levels using Safe-SeqS prior to cycle two chemotherapy was associated with objective radiological response at 8–10 weeks (p = 0.016) [73]. This study also found a trend between ctDNA reduction and improved progression-free survival. The more recent PLACOL study in 82 patients receiving chemotherapy for mCRC echoed these findings [7]. PLACOL utilised picodroplet-digital PCR assays based on either genomic or hypermethylation alterations. The investigators found that the baseline ctDNA concentration was prognostic for OS, and that early and deep ctDNA reductions were associated with improved objective response rate and longer survival (p < 0.001) [7]. Another recent study using digital PCR found methylation changes over time correlated with

These studies suggest that early changes in ctDNA during systemic therapy may be predictive for treatment efficacy and prognostic for survival outcomes, thus suggesting a role for serial ctDNA monitoring during palliative treatment with systemic therapy. Indeed, with the advantages of a short half-life reflecting immediate-term changes [18] and high tumour-related specificity, ctDNA monitoring may be complementary to radiological assessments and blood biomarkers currently in use. In clinical circumstances where radiological assessments are indeterminate or ambiguous, such as the lack of measurable disease by imaging criteria or the presence of mixed response, ctDNA dynamics may be of particular value; although ctDNA may not always correspond to imaging findings [75]. Equally, it is also prudent to acknowledge that no current evidence supports the strategy of biomarkermonitoring of palliative therapy and that earlier adaptive treatments will augment

**5.2 CtDNA in mCRC: genotyping to guide targeted treatment selection**

It is now standard of care for patients with mCRC to undergo molecular profiling on their tumour tissue in order to determine *BRAF V600E* and *Rat sarcoma* (*RAS*, particularly exon 2–4 *KRAS*) gene mutational status. This informs clinical decision-making regarding benefit from anti-EGFR therapy. The advantageous role of genotyping with ctDNA has already been established in the field of advanced non-small cell lung cancer. Circulating genetic aberrations of EGFR (for example, exon 19 deletions or exon 21 (L858R) substitution mutations) are now being used in standard practice, as a companion tool, to identify eligible patients for treatment with erlotinib. This technology was approved by the US Food and Drug

In mCRC, a meta-analysis of 21 studies on the diagnostic performance of ctDNA-based *KRAS* gene detection found overall high pooled specificity (96%)

**62**

and moderate sensitivity (67%) [77]. Not surprisingly, heterogeneity was noted to be high probably owing to varying molecular techniques, tumour stage and study designs [77]. Although a number of *KRAS* ctDNA assays have demonstrated high agreement (91–93%) with tumour tissue *KRAS* testing and are available for commercial use [78, 79]; at present, tumour tissue testing remains the gold standard to establish *KRAS* mutational status. Given the appreciable discordance rate with tumour tissue genotyping, it is recommended that a negative ctDNA result should trigger tumour tissue variant analysis [5]. As will be discussed under Section 6, ctDNA assays in mCRC may also been utilised to select predictive immune-related biomarkers for immunotherapy selection.
