Reflecting the high incidence of colorectal cancer, CEA is one of the most commonly investigated and used tumour markers [9]. A foetal glycoprotein involved in cell adhesion, CEA is one of the first tumour biomarkers, identified in 1965 by Gold and Freedman [10]. It is normally produced by the epithelium of the large intestine of the foetal gastrointestinal tract, but CEA is detectable in very low serum levels of healthy adult individuals.

Raised CEA expression is associated with many cancers including breast, respiratory, genitourinary and gastrointestinal cancers [11]. The action of CEA has been shown to be central in the multistep cascade by which malignant cells colonise the liver, including apoptosis inhibition [12], facilitation of malignant cell adhesion [13] and evasion from Kupffer cell detection [14]. Unsurprisingly therefore, high serum levels of CEA in patients with CRC are associated with the development of liver metastases [15].

Following curative-intent surgery for primary CRC, numerous studies have demonstrated that patients following an intensive surveillance regime that includes regular CEA testing had a significantly favourable outcome than patients undergoing surveillance without CEA testing [16]. As such, most expert panels in North America and Europe recommend serial measurements of CEA following curative-intent surgery for CRC [17, 18, 19].

In the case of follow-up for liver metastasis, following resection of curative intent, CEA measurement is recommended at 3–6 month intervals for 3 years after surgery [20]. However, CEA measurement alone is inadequate. A series following 314 patients reported that recurrences were detected in 23% through a CEA increase alone without any relevant findings on routine imaging, in 46% through a CEA rise with simultaneous positive imaging, and in 31% by positive imaging alone [21].

A significant elevation is considered to be at least 30% over that of the previous elevation whilst the recommendation is that CEA should be measured 3 monthly for 3 years following metastasectomy [18]. Elevated CEA levels should prompt further investigation in the form of imaging to detect potential metastatic disease.

Although more commonly associated with pancreatic and biliary tumours, elevated serum CA 19-9 levels have also been linked to CRC. CA 19-9 is a tetrasaccharide carbohydrate also known as sialyl Lewis A and is synthesised by gastrointestinal epithelium. Paradoxically, it was first discovered to be elevated in colon cancer [22], however, due to its low sensitivity and specificity it has not been established as a reliable tumour marker in this setting.

Some studies have shown that elevated CA 19-9 levels have been linked with poorer prognosis [23]. A recent study by Thomsen et al. [24], as part of the NORDIC VII trial, investigated the prognostic factor of CA 19-9 in patients with unresectable metastatic colorectal cancer and RAS or BRAF mutations. A total of 566 patients were investigated and elevated CA 19-9 (and CEA) levels were associated with a reduced overall survival, most notably in CRC expressing the BRAF mutation.

We did not identify any studies suggesting a role for CA 19-9 in following up patients with resected primary colon cancer or within the setting of resected CRC liver metastases.

CA 72-4 is an antigen to tumour-associated glycoprotein 72 (TAG-72), a 48-kDa mucin-like glycoprotein complex first identified in 1981 [25]. CA 72-4 has been found in a variety of human adenocarcinomas including colorectal, gastric, breast, lung and ovarian, but is rarely expressed in normal or benign tissue [26]. In a study of 106 patients (53 colon and 53 rectal) Sing et al. [27] found a diagnostic sensitivity of 45.3% and specificity of 95.9% in 53 patients with colon cancer, and 69.4% and 63.9%, respectively, in rectal cancers. However, when used in combination with other tumour markers (TK1, CEA and CA 19-9) sensitivity increased to 86.9% and 96.0% for colon cancer and 80.9% and 96.1% for rectal cancer. Carpelan-Holmström et al. [28] studied various tumour markers including CEA, CA 19-9 and CA 72-4 in 102 patients with resected colorectal cancer, in particular their value in detecting recurrence. Only CEA proved to be of diagnostic significance. We did not identify any studies justifying the use of CA 72-4 in the surveillance of resected CRC liver metastases.

The DR-70 test marketed under the brand Onko-Sure is a serum ELISA assay that quantifies fibrin and fibrinogen degradation products in blood. These have been found to be elevated in various cancers including CRC. DR-70 screening is approved for use in many countries including the US and Europe for lymphoma and lung, breast, stomach, liver, ovarian, oesophageal, cervical, trophoblastic, thyroid, brain and pancreatic cancers, as well as CRC. It was approved by the USFDA in 1982 and was the first test cleared for monitoring of CRC after CEA. Although it does not feature in societal guidelines, there is some evidence that it may be useful in patients who have a positive histological diagnosis of CRC with normal CEA levels [29].

There is however no published evidence of any role in detection of recurrence after liver resection.

Matrix metalloproteinases (MMPs), also referred to as matrixins or matrix metallopeptidases are enzymes capable of breaking down extracellular matrix proteins. First described in 1962 [30], they have been shown to play a major role in physiological and pathological cell activities such as cell proliferation, migration (adhesion/dispersion), differentiation, angiogenesis and apoptosis [31]; it therefore comes as no surprise that, they have a major role in cancer development and spread [32] in addition to being a potential biomarker of malignancy [33].

As their name suggests, TIMPs, are a family of intrinsic inhibitors of MMPs and play an important role in the regulation of MMP activity. An imbalance between TIMPs and MMPs can result in an altered extracellular matrix in turn, leading to early tumour development and metastasis [34]. Several studies have reported this, including ones specifically investigating CRC [35, 36].

Huang et al. [37] recently investigated the serum levels of various biomarkers including several MMPs in patients with CRC. This case-controlled study included 112 patients with newly diagnosed CRC at various stages. All cases were confirmed with positive histopathology and were compared to 115 matched controls. They demonstrated that serum MMP-7, MMP-9, MMP-11, TIMP-1, TIMP-2, CEA, and CA19-9 levels were all significantly higher than in the control group.

More recently, Reijonen et al. [38] published a study with 419 patients who underwent hepatectomy for CRC metastatic disease. The group investigated the prognostic value of various serum biomarkers including MMP-8 and MMP-9. Pre- and postoperative elevated MMP-8 levels were associated with a worse overall survival at 10 years, whilst MMP-9 did not serve as a prognostic marker.

No studies specifically looking at the diagnostic value of MMPs detected in patients with resected liver metastases from colorectal cancer were identified.

MiRNAs are a vital component for the non-coding RNA family, that help to shape the expression of most mRNAs. Hundreds of different miRNAs have been isolated in humans and are thought to influence all developmental disease and processes [39]. By binding to tumour suppressor genes or oncogenes, miRNAs regulate a number of oncogenic processes including cell growth, apoptosis, cell differentiation and angiogenesis [39]. MiRNA dysregulation was first reported in chronic lymphocytic leukaemia [40] and has since been associated with various cancers, including CRC.

A wide range of miRNAs have been found to be either up or/and downregulated in patients with CRC [41]. The first report came from Michael et al. [42] in 2003 reporting that miR-143 and miR-145 were downregulated in CRC tissue samples compared to normal tissue. Further investigators have reported that patterns of tissue miRNA expression can be associated with prognosis and chemotherapy resistance [43].

Of further interest, however, is that circulating miRNAs found in blood may also be of diagnostic or prognostic value. Han et al. [44] used a panel of four miRNAs, namely miR-15b, miR-16, miR-21, miR-31 to create a panel test and reported a diagnostic sensitivity and specificity of 95.06% and 94.44%, respectively, when comparing healthy individuals to patients with CRC. The same panel reported a sensitivity and specificity of 85.19% and 82.09%, respectively, when comparing patients with CRC with patients having colorectal adenomas. Huang et al. [45] reported a diagnostic sensitivity of 81.25% and a specificity of 73.33% using a panel of four alternative miRNAs: miR-203a-3p, miR-145-5p, miR-375-3p and miR-200c-3p.

A meta-analysis of 18 studies that included over 2000 patients and controls showed a pooled sensitivity and specificity for circulating miR-21 of 77% and 83%, respectively [46].

Nassar et al. [47] investigated a cohort of patients with stage IV CRC to identify miRNA panels that could be used for diagnosis of Stage IV CRC in general, and liver metastasis in particular. They reported that the combination of miR-210 and miR-203 together offers a diagnostic accuracy of 72% for patients with CRC and liver metastases. Nevertheless, despite the above advances, miRNA testing has yet to enter in routine clinical use as a biomarker either in the setting of primary colorectal cancer or in the detection of recurrence after liver resection.

Several recent reports have championed the potential use of detecting circulating tumour DNA (ctDNA) in the management of patients with cancer [48]. Although free DNA was first reported in blood in 1948 [49], the categories of body fluids that can be profiled has expanded since to include urine, saliva, cerebrospinal and pleural fluid. As circulating free DNA (cfDNA) is mostly derived from cells undergoing necrosis or apoptosis, higher levels are found in patients with cancer, in addition to insults such as trauma [50], infraction, transplant rejection and infection [51]. The fraction of cfDNA originating from tumour cells is termed ctDNA and along with detection of CTCs is often referred to as a liquid biopsy [52].

Liquid biopsies are broadly categorised into whole genome sequencing and PCR-based tests targeting specific genes. In principle, ctDNA should contain the same genetic defects as the original tumour cells [52]. Detectable Ct DNA has therefore been used to detect mutations and guide treatment such as epidermal growth factor receptor (EGFR) therapy in lung cancer and Ras in colorectal cancer [53, 54]. In addition, commercial products that perform whole genome DNA sequencing from liquid biopsy have been licenced by in the US, Japan and EU. Products such as FoundationOne Liquid CDx and Guardant360 CDx [55, 56] can perform comprehensive tumour genome profiling from liquid biopsy and provide information about genomic alterations to guide the use of approved targeted therapies and ongoing clinical trials [57].

The CIRCULATE trial, currently recruiting patients, is investigating the role of ctDNA in patients with stage II colorectal cancer and aims to establish if detectable ctDNA can guide adjuvant treatment. As a secondary objective however, it will investigate the potential of ctDNA as a follow-up tool in resected patients [58].

Bolhuis et al. [59] investigated 23 patients who had already been enrolled in an ongoing phase 3 trial (the CAIRO5 study) and who had resected colorectal metastatic disease to the liver. This was the first and, to date, only study to analyse the association of postoperative ctDNA in the setting of resected CRC liver metastases. CtDNA analysis was performed at baseline, prior and post liver resection and showed that postoperative ctDNA status is an independent prognostic factor for recurrence. Although this is a small study, the authors suggest that it is a promising biomarker for disease monitoring post liver resection.

To date there have been no trials specifically looking at the diagnostic role of ctDNA detection in patients with resected liver metastases from colorectal cancer, however, this avenue has gained traction and appears to be a promising direction.

EVs are cell membrane-derived vesicles which are actively secreted by cells. They are broadly subcategorised as microvesicles and exosomes depending on their size, and are a heterogeneous entity; they mediate cell signalling in both physiological and pathological settings. As they are detectable in all body fluids, including blood, they have been lauded as a potential biomarker of disease, including cancer [60]. In addition, they have been considered a therapeutic delivery vehicle or as a target themselves to inhibit pathological cell signaling [61].

EVs are nanosized and comprised of a lipid bilayer surface that carries bioactive cargo in the form of proteins of genetic materials such as DNA, mRNA or miRNA and they form a snapshot of the secreting parent cell [62]. Investigation of EVs involves isolation of EVs from a body fluid, followed by composition analysis [63]. Their relative abundance in blood and stability due to the protective membrane have made EVs of particular interest as biomarkers in many cancers [64].

Silva et al. [65] analysed plasma from 91 patients with colorectal cancer. They found that, when compared to blood derived from healthy individuals, patients with colorectal cancer had a statistically significant higher number of detectable exosomes. However, within the cancer group, patients with high levels of exosomes had a similar length of disease-free survival, although they had a shorter overall survival when compared to patients with lower number of exosomes.

There are no substantial studies on human patients specifically looking at the biomarker role of EVs in patients with resected liver metastases from colorectal cancer. Shao et al. [66] investigated the role of EVs in promotion of liver metastasis in a murine model and found that miR-21 containing EVs enable this by creating a proinflammatory state by enhancing IL-6 production. As part of this study, plasma from 25 patients with CRC and liver metastases was compared with those of 20 patients without liver metastases, 35 healthy patients and 29 patients with adenomas. EV miR-21 was detected in increasing levels in patients with higher tumour stage.

It is widely acknowledged that the modification of gene expression plays an integral role in tumorigenesis and cancer progression. Hence, the search for epigenetic markers of cancer has been another route under investigation for reliable biomarkers in many cancers, including CRC [67]. Although histone modifications have been studied, most of the focus has been on DNA methylation biomarkers [68].

The SEPT9 gene, coding for the Septin-9 protein, has been shown to be methylated in colorectal cancer when compared to normal colon mucosa. The PRESEPT trial was a large prospective study of nearly 8000 asymptomatic patients which showed that screening for CRC by means of detecting circulating methylated SEPT9 DNA (mSEPT9) resulted in a sensitivity of 48.2% and specificity of 91.5% [69]. This led in 2016 to European and FDA approval of the Epi proColon® screening test, the first methylation-based biomarker for CRC [70, 71, 72]. It is only indicated, however, in patients above the age of 50 who have an average risk of developing CRC and who cannot or wish not to partake in regular methods of screening.

There is no licence for the use of this test for monitoring patients with resected primary or metastatic CRC. However, recent studies have shown that mSEPT9 levels corelate well with treatment response undergoing hepatectomy. Liu et al. [73] studied 51 patients who had had either synchronous or metachronous hepatectomy for CRC with curative intent. The group reported that mSEPT 9 levels dropped considerably following liver resection and patients with higher levels had a poorer prognosis. However, mSEPT9 levels were not used to follow up patients with respect to early detection of recurrence.

Other DNA methylation biomarkers under investigation include Twist-Related Protein 1 (TWIST1) [74], Runt-Related Transcription Factor 3 (RUNX3) [75], Tachykinin-1 (TAC1) [76], Insulin-Like Growth Factor Binding Protein 3 (IGFBP3) [77], Eyes Absent Homolog 4 (EYA4) [78] and Somatostatin (SST) [79], however, none of these has been as widely investigated as SEPT9.