**3. CtDNA in screening**

*Advances in the Molecular Understanding of Colorectal Cancer*

accuracy, while minimising cost [5].

tions of DNA [19, 20].

analysed.

routine clinical practice in a credentialed laboratory would require considerable scaling up, standardisation, and optimisation of methodological efficiency and

Earlier approaches such as Sanger sequencing and conventional polymerase chain reaction (PCR)-based methods have limited sensitivity for ctDNA detection, particularly for rare alterations [6]. A number of digital PCR-based approaches were subsequently developed, capable of improved limits of detection (up to 0.001%), low frequency allele detection and nucleic acid quantitation. Commonly used digital PCR methods for ctDNA detection include BEAMing (beads, emulsion, amplification and magnetics) and droplet digital PCR (ddPCR). BEAMing which combines beads in emulsion and flow cytometry, was first described in 2003 [17] and facilitated serial tracking of mutant allele fractions in patients with CRC [9, 18]. This method and its variations are now widely applied in ctDNA analysis. DdPCR involves the massive partitioning of nucleic acids into thousands of droplets to enable highly sensitive and precise detection and quantitation of small concentra-

Next generation sequencing (NGS) or massively parallel sequencing are broad terms describing a range of high throughput methods capable of the simultaneous analysis of thousands to millions of DNA molecules, and also encompasses both targeted and non-targeted approaches. Targeted sequencing platforms such as safe sequencing system (Safe-SeqS) [21] and tagged-amplicon deep sequencing (TAm-Seq) [22] have the advantage of improved multiplex capability and evaluating a larger number of loci simultaneously in the genomic areas of interest [15]. However, targeted PCR-based or NGS methodologies mostly rely on prior knowledge of molecular changes and cannot identify variants located in areas that are not

In contrast, non-targeted genome or exome-wide sequencing allows discovery of *de novo* alterations as well as detection of structural changes including rearrangements, gene fusions and copy number alterations [23]. These may be advantageous in patients who do not have accessible tumour tissue for biopsy. Several techniques have been described and used in the ctDNA setting. The personalised analysis of rearranged ends (PARE) method uses paired-end sequencing, and was utilised in a proof-of-principle study to identify unselected genome-wide chromosomal alterations characteristic of tumour DNA in cfDNA in patients with CRC and

**Technique types Technique Application**

PCR Bi-PAP

DdPCR

Safe-SeqS TAm-Seq CAPP-Seq

Digital karyotyping

*genome sequencing; WES, whole-exome sequencing; PARE, personalised analysis of rearranged ends.*

*PCR, polymerase-chain reaction; ARMS, amplification-refraction mutation system; Bi-PAP, bidirectional pyrophosphorolysis-activated polymerisation; BEAMing, beads, emulsion, amplification and magnetics; ddPCR, droplet digital PCR; Safe-Seq, safe-sequencing system; TAm-Seq, tagged-amplicon deep sequencing; WGS, whole-*

Mutant allele-specific

Known point mutations

Known point mutations

Copy number variations Structural rearrangements

Structural changes in specific gene

Point mutation

regions

PCR-based [26–28] ARMS-PCR

Digital PCR [18–20, 29] BEAMing

WGS and WES [24, 31, 32] PARE

Targeted sequencing [21, 22, 30]

*Methods of ctDNA detection.*

**58**

**Table 1.**

Five-year survival for CRC patients is highly dependent on the timing of disease detection and tumour stage. CRC screening can achieve early disease detection and treatment, including that of pre-malignant dysplastic lesions, and has been shown to improve CRC-related mortality. However, 60–70% of patients are diagnosed at mid- to late stage CRC despite recent advances in screening methods [33]. Screening approaches used to test asymptomatic people for a presence of unsuspected disease, which have proven efficacy in CRC include endoscopic visualisation and faecal occult blood tests (FOBT). The former is invasive and expensive with associated morbidity, thus affecting patient compliance and acceptance. FOBT measure the presence of haemoglobin in faeces and can be categorised into guaiac-based (gFOBT) and the newer haemoglobin-antibody-based faecal immunochemical test (FIT). FIT is used more commonly in current practice and has largely superseded gFOBT, due to its superior sensitivity for colorectal bleeding, improved analytical characteristics and it is also less likely to be affected by dietary and medication factors [34–37]. In addition, FIT has better acceptance and participation which improves population participation [38]. In a meta-analysis of 19 studies, FIT was found to have high accuracy and specificity, and moderately high sensitivity, although substantial heterogeneity was noted across studies [39].

This prompted the development of faecal-based tests targeting genetic and epigenetic alterations. Cologuard is the first commercially-approved faecal test which combines several technologies including molecular assays for aberrant *NDRG4* and *BMP3* methylation, *β-actin* (a reference gene for human DNA quantity), and *KRAS* mutations; and a haemoglobin immunoassay [40]. The haemoglobin component of the Cologuard test contributes to 80% of the cancer detection in the algorithm. A large randomised clinical trial comparing Cologuard and FIT screening showed that the sensitivity of Cologuard was superior to that of FIT in the detection of CRC and precancerous lesions [40]. However, the higher cost of Cologuard and its lower specificity compared with FIT has limited its adaptation as a screening tool [33].

CtDNA analysis may offer a more convenient screening approach compared with faecal-based tests. The malignant transformation pathway, from adenoma to carcinoma, is driven by mutations such as *APC*, *KRAS* and *TP53* [41]. However, somatic mutational profiles are highly variable between patients. For example, *KRAS* and *BRAF V600E* are seen in approximately 40 and 7% of patients with CRC respectively [42]. To date, the vast majority of cancer patients evaluated with mutation-based blood plasma assays have advanced-stage disease. A challenge in early stage disease is the often-minute fraction of ctDNA present in the total circulating DNA—may be as low as <0.01%—which may be below the limit of detection assays [18]. A study which enrolled 170 patients with positive FOBT investigated differences in *KRAS* mutation levels in plasma and tissue samples [43]. The rate of *KRAS* mutations in plasma (3%) was found to be lower compared with

that observed in matched adenocarcinoma and high-grade intra-epithelial neoplasia tissues (45%) [43]. Although this is a small study, the results suggest that either this particular assay is not sufficiently sensitive, or that ctDNA was found at low or undetectable levels in the population tested.

The detection of epigenetic alterations characterised by aberrant DNA methylations is an alternative approach to mutational ctDNA analysis. Aberrant DNA methylation leads to transcription silencing of tumour suppressor genes, occurs early in CRC carcinogenesis, and may be more commonly seen and consistent in cancer patients compared with somatic mutations. Indeed, DNA methylation profiles in plasma have been used in biomarker development to identify emergence of early CRC by several groups [44]. One of the methylated promotors that has gained a lot of interest is the *Septin 9* gene promoter. Methylation in the *Septin 9* promoter demonstrated high sensitivity in preclinical studies and a small clinical cohort [45, 46], however a large prospective screening study demonstrated a sensitivity profile of only 48% [47]. In addition, the sensitivity to detect advanced adenomas was low (11%) [47]. Despite this, in 2016, the FDA approved the use of the Epi pro-Colon, a plasma-derived *Septin 9* methylation assay, for the screening of CRC. This decision occurred in the setting of encouraging results from a meta-analysis comparing the pooled sensitivity of methylated *Septin 9* with FOBT as a screening tool [48], and the improved sensitivity and specificity results of a modified version of the Epi proColon assay (2.0 version) [49, 50]. Recently, promising results have been reported in the utilisation of two methylation markers in the screening context—*branched-chain amino acid transaminase 1* (*BCAT1*) and *ikaros family zinc finger protein 1* (*IKZF1*)—where methylation of either gene identified close to 70% of CRC with specificity of 92% [51, 52].

To date, ctDNA is yet to demonstrate clinical utility in the CRC screening setting. Challenges lie in minimising false positive readings, whilst developing a test sensitive enough to detect small amounts of ctDNA. For example, normal physiological ageing is associated with the development of somatic mutations in the absence of malignant disease, and false positive readings may also be seen in patients with chronic inflammatory disease. False positive results can lead to unnecessary follow-up procedures and anxiety. Studies examining a large number of healthy control individuals will be essential to evaluate the specificity of potential screening assays. Yet another challenge with ctDNA-based screening is the identification of the underlying tissue of origin. Because the same gene mutations drive multiple tumour types, ctDNA tests based on genomic analysis alone generally cannot identify the anatomical location of the primary tumour.
