The biological complexity of colorectal cancer: insights into biomarkers for early detection and personalized care

Colorectal cancer: epidemiology and therapy

Colorectal cancer (CRC) is one of the main causes of cancer deaths and the third most frequent cancer worldwide, with about 30% of cases being hereditary or familial [Ferlay et al. 2010, 2015; Karsa et al. 2010]. There are several different causes involved in CRC incidence. Sporadic CRCs arise following somatic mutations, and these account for around 75% of all CRCs. Germline-inactivating mutations in oncogenes or tumor suppressor genes cause hereditary CRC, while minor variant or single nucleotide polymorphisms in the same genes are responsible for familial CRCs. First-degree relatives of patients with CRC have a three-fold greater risk of CRC than that of individuals without familial predisposition [Jasperson et al. 2010; Duraturo et al. 2011; Rustgi 2007; De Rosa et al. 2015].

Hereditary nonpolyposis CRC (HNPCC), or Lynch syndrome, is an autosomal dominant disease that is caused by a germline mutation in one of the DNA mismatch repair (MMR) genes, such as MLH1 and MSH2, or also MSH6 and PMS2. HNPCC is the most common form of hereditary CRC, and it accounts for 1–6% of all colorectal malignancies [Aaltonen et al. 1998; Samowitz et al. 2001]. Patients with HNPCC should undergo colectomy with ileorectal anastomosis, while carriers of a mutation in these MMR genes should undergo colonoscopy every 1–2 years from the age of 20–25 years, and for woman, pelvic evaluation should be followed from the age of 25–35 years. Also, relatives of these patients should be tested for MLH1 and MSH2 mutations [Akiyama et al. 1997; Bronner et al. 1994; Fishel et al. 1993; Miyaki et al. 1997; Nicolaides et al. 1994; Niessen et al. 2009].

Adenomatous polyposis and hamartomatous polyposis are inherited syndromes that account for 2–5% of all colon cancers. Adenomatous polyposis syndromes include familial adenomatous polyposis (FAP), attenuated FAP (AFAP), and MUTYH-associated polyposis (MAP). Hamartomatous polyposis syndromes include Peutz–Jeghers syndrome, juvenile polyposis syndrome, and ‘PTEN hamartoma tumor syndrome’. AFAP and MAP, and also Lynch syndrome, can all be associated with only a few adenomas (i.e. 3–10 polyps); nevertheless, each syndrome has distinct cancer risks, characteristic clinical features, and separate genetic etiology. Thus, differential diagnosis is essential for correct management of the specific disease [Lucci-Cordisco et al. 2013]

Familial adenomatous polyposis is a hereditary autosomal dominant syndrome that is mainly associated with mutation of the tumor suppressor APC (adenomatous polyposis coli) gene, which is located on chromosome 5q21-q22 [Groden et al. 1991; Kinzler et al. 1991]. FAP is characterized by the onset of hundreds to thousands of adenomas at the level of the large intestine, which usually occurs at a young age. If these are untreated, progress is almost invariably (90%) toward the development of CRC at an average age of 40 years [Vasen et al. 2008]. Patients who carry an APC mutation should undergo colonoscopy every 1–2 years from the age of 10–12 years, and in the presence of adenomatous polyps, they need, instead, annual endoscopic follow up. Prophylactic colectomy is recommended if the polyps are endoscopically untreatable, and proctoscopy should be performed every 6 months, with monitoring for ileum–anal anastomosis every 2 years if total proctocolectomy has been performed [Jasperson et al. 2010].

Patients with ulcerative colitis and Crohn’s disease are associated with high risk of developing CRC. According to a relatively recent meta-analysis, the risk here is estimated at 2%, 8%, and 18% after 10, 20, and 30 years, respectively. The duration and extent of the ulcerative colitis or Crohn’s disease are the two most important factors that determine the patient cancer risk. Patients with irritable bowel disease who have a family history of CRC have double the cumulative risk of developing dysplasia, compared with those with a negative history [Eaden et al. 2001].

The trend in the incidence of CRC by age follows the typical curve of the majority of cancers, increasing with age. Indeed, CRC diagnoses are sporadic in the younger age groups, while the frequencies are of the order of 50 cases per 100,000 inhabitants/year between 30–50 years, which increases to 100 cases between 50–70 years, and reaching about 250 cases at ages >70 years. The incidence rates are similar between the two sexes up to the age of 45 years; however, the curve for males then shows a steeper increase and reaches higher values.

Surgery is the mainstay of treatment for early colorectal tumors. Colon cancers are treated by surgical excision and with complete mesocolic excision [Hohenberger et al. 2009; Sehgal and Coffey 2014] which appears to be as appropriate as total mesorectal excision for rectal cancer. These allow excellent oncological outcome, with a 5-year cancer-specific survival rate of 91.4% in stage II, and 70.2% in stage III [Siani and Pulica, 2014]. In the case of obstruction, perforation, and bleeding, emergency segmental colectomy for resection of the tumor is indicated [Sagar, 2011]. Several international multicenter randomized trials have shown that minimally invasive surgery is a safe and valid alternative to open surgery, with less postoperative pain, shorter duration of hospital stay, and similar oncological outcomes. Robotic surgery has not been shown to be better for either colon or rectal cancer, compared with an adequate laparoscopic approach [Tyler et al. 2013; Baek et al. 2013].

The staging of rectal cancer is crucial to define the most appropriate treatment strategy, with some important factors defined here, such as distance from the anal verge, circumferential margin, nodal stage, and vascular and nerve invasion. Local excision can be considered for very early tumors (i.e. stage 1) and for malignant polyps with an invasion level of ‘0–3’; this should also be supported by accurate preoperative diagnosis with endorectal ultrasound, magnetic resonance imaging, and computed tomography scanning. A transanal procedure as a video-assisted technique can be performed, with either transanal endoscopic microsurgery [Sgourakis et al. 2011; Guerrieri et al. 2014] or transanal minimally invasive surgery [Matz and Matz, 2012; Atallah et al. 2010]. In the more advanced disease stages, surgery with total mesorectal excision is the appropriate procedure, either performed laparoscopically or as open surgery. Neoadjuvant chemoradiotherapy is indicated in locally-advanced rectal cancer, to reduce local recurrence rates. This became standard practice after the publication of the results of the German CAO/ARO/AIO-94 randomized phase III trial [Sauer et al. 2012]. Long-course chemoradiation and short-course radiation can both be considered here [Sauer et al. 2012; Schmoll et al. 2012]. Infusional 5-fluorouracil has been commonly used in neoadjuvant treatment, although the oral capecitabine appears to be equally effective and more manageable [Hofheinz et al. 2012].

Stage IV disease (i.e. CRC with metastatic sites, mCRC) is commonly treated with chemotherapy, with several drugs used in combination with 5-fluorouracil and leucovorin (e.g. oxaliplatin–FOLFOX, irinotecan–FOLFIRI). There are two monoclonal antibodies against the epidermal growth factor receptor (EGFR) that are now commercially available (i.e. cetuximab and panitumumab) [Downward 2003], and these are used in combination with commonly used treatment schedules [Tveit et al. 2012; Peeters et al. 2010; van Cutsem et al. 2009]. The National Cancer Institute of Canada Clinical Trials Group CO.17 trial demonstrated that the anti-EGFR monoclonal antibody cetuximab improves overall survival (OS) and progression-free survival (PFS) in patients with advanced, chemotherapy-refractory CRC, particularly in patients with wild-type KRAS tumors [Au et al. 2009].

These therapies show very low objective response rates because of the high Ras/Raf mutation rates in CRC [Bazan et al. 2002]. Biological chemotherapy also includes the vascular endothelial growth factor (VEGF)-A-targeted antibodies bevacizumab and aflibercept, recombinant proteins that target VEGF-A, VEGF-B, and placental growth factor (PlGF), and these have been approved by the US Food and Drug Administration for mCRC, in combination with standard cytotoxic chemotherapy [Ciombor et al. 2015; Hurwitz et al. 2004; van Cutsem et al. 2012; Tabernero et al. 2014]. New therapeutic agents in the targeting of the EGFR and VEGF receptor (VEGFR) pathways, and those targeting other signal-transduction pathways, such as the MET, IGF1R, MEK, phosphoinositide 3-kinase (PI3K), Wnt, Notch, Hedgehog, and death-receptor signaling pathways, are now under evaluation for the treatment of mCRC [Seow et al. 2016].

Some groups claim that regular use of nonsteroidal anti-inflammatory drugs (NSAIDs), including aspirin, has an important role in the prevention of CRC; according to some studies, these can reduce the risk by 40% [Cuzick et al. 2009]. Furthermore, randomized studies have evaluated the preventive effects of aspirin on the development of adenomatous polyps, with reports of an inverse association between use of NSAIDs and incidence of CRC [Bosetti et al. 2006, 2009; Chan et al. 2005; Dubé et al. 2007]. Although numerous studies have shown preventive effects of aspirin on incidence and size of adenomatous polyps [Gann et al. 1993; Baron et al. 2003; Benamouzig et al. 2003; Chan et al. 2004], the dose and duration of treatment have never been specified, and some studies have failed to demonstrate effects of aspirin on the incidence of any specific cancer [Cook et al. 2005]. More randomized trials are mandatory to clarify the real preventive role of aspirin in CRC.

Model for colorectal cancer tumorigenesis

The development of CRC is characterized by the ‘adenoma–carcinoma sequence’. Thus, the normal epithelium acquires sequential genetic and epigenetic mutations in specific oncogenes or tumor suppressor genes, becomes a hyperproliferative mucosa, and later gives rise to a benign adenoma that changes into a carcinoma and metastases over an average of 10 years (Figure 1) [Pancione et al. 2012; Ewing et al. 2014; Vogelstein et al. 1988].

OPEN IN VIEWER

Figure 1.

Colorectal cancer heterogeneity during the adenoma carcinoma sequence.

Normal gastrointestinal epithelium is organized along a crypt–villus axis. At the bottom of the crypts, there is a pool of colon stem cells and progenitor cells, which are the most undifferentiated cell types that can undergo self-renewal and show pluripotency. These cells can move along the crypt–villus axis while differentiating into all of the epithelial colon lineages, such as Paneth cells, goblet cells, enterocytes, and enteroendocrine cells. In about 14 days, they arrive at the top of the villus, and undergo programmed cell death by apoptosis [Peifer, 2002; Kosinski et al. 2007]. This process is orchestrated by gradients of proteins, such as Wnt, bone morphogenetic protein (BMP) and transforming growth factor (TGF)-β, together with the extracellular matrix and stromal cells that form the cell niches [Medema and Vermeulen, 2011].

The stem cell niches thus have a critical role in the maintenance of stem cell properties, by providing them with protection from differentiation and apoptosis, and through controlling their turnover [Moore and Lemischka, 2006; Medema and Vermeulen, 2011]. It has been suggested that these stem cells are involved in cancer onset and progression, through their acquiring tumorigenic potential, and thus transforming into cancer stem cells (CSCs). These CSCs represent the only tumor cells that can initiate cancer transformation, progression, and metastases, and they are probably responsible both for relapse and resistance to therapy [Gangemi et al. 2009; Fanali et al. 2014]. The source from which CSCs arise is still unclear. It remains a topic for discussion whether differentiated cancer cells can de-differentiate, and thus transform into CSCs, or alternatively, whether healthy stem cells accumulate genetic mutations that confer them a neoplastic phenotype [O’Brien et al. 2007; Greaves and Maley, 2012]. Moreover, in contrast to the clonal hypothesis that was formulated to explain the mechanisms responsible for cancer onset and progression, the CSC hypothesis is a hierarchical theory according to which only a few undifferentiated cells, the CSCs, can promote tumor onset, progression, and metastases [Puglisi et al. 2009]. In agreement with this hypothesis, tumors should be composed mainly of differentiated cancer cells and of a few undifferentiated CSCs, which although slowly duplicating, are the only cells that can undergo self-renewal and show multipotency [Vermeulen et al. 2008].

It has been demonstrated recently that during epithelial to mesenchymal transition (EMT), epithelial cancer cells acquire not only mesenchymal traits, but also a stem cell-like phenotype, with the ability for self-renewal, unlimited proliferation, and resistance to apoptosis. This observation suggests that EMT might represent a tumor source of undifferentiated cancer cells, as the CSCs or stem cell-like cells [Dallas et al. 2009; Fan et al. 2012]. EMT consists of an essential phenotypic conversion of epithelial cells into cells with a mesenchymal phenotype. It is a reversible process that often occurs during embryonic development and tissue remodeling, and it also has a critical role in the early events that occur in invasion and metastases of many types of cancer, including CRC [Batlle et al. 2000]. EMT regulation is orchestrated through a group of transcription factors, including Snail, Slug, ZEB1 and Twist, although the tumor microenvironment also has a role in this conversion through different signals, such as TGF-β, EGF, Wnt and Notch [Cano et al. 2000; Li et al. 2014].

During EMT, epithelial cells lose E-cadherin expression, which specifically guarantees the epithelial phenotype, with destruction of their intercellular adhesion, and acquisition of mesenchymal characteristics and increased migratory and invasive properties [Thiery et al. 2009]. Furthermore, the EMT program induces stem cell-specific gene expression, thus promoting the self-renewal capability [Mani et al. 2008].

Metastases are the first cause of death in CRC, as well as with many other solid tumors, and this defines stage IV CRC, which is characterized by relatively short OS [Singh and Settleman, 2010]. During metastases development, the cells acquire the ability to invade the basal membrane that divides the epithelial tissues from the extracellular matrix, and they migrate into surrounding tissues. Later, they join the endothelial cells from lymphatic and blood vessels and arrive in the lumen, in a process known as intravasation. Subsequently, the cells that survive in the lumen of the vessels come out from the vases (i.e. extravasation), disseminate into the adjacent organs, and finally colonize them to generate micrometastases,

Thus, as described above, EMT confers to the epithelial cells the traits that are needed to enable them to complete this process [Singh and Settleman, 2010; Thiery et al. 2009].

Molecular heterogeneity of colorectal cancer

At the molecular level, CRCs are a very heterogeneous group of diseases, and this appears to have a critical role in the generation of drug resistance, and ultimately, in the mechanism of relapse. Loss of genomic integrity facilitates accumulation of multiple mutations during the development of CRC. Chromosomal instability, microsatellite instability (MSI), aberrant DNA methylation (i.e. the CpG island methylator phenotype), and DNA repair defects are all mechanisms that are involved in colorectal epithelial cell transformation, and they all have a significant role in CRCs (Figure 1) [Goel et al. 2007; Sideris and Papagrigoriadis, 2014; Pino and Chung, 2010; Boland and Goel, 2010; Colussi et al. 2013]. It is well accepted that a tumor is classified as positive in terms of the CpG island methylator phenotype if it shows methylation of at least three of the following markers: CACNA1G, IGF2, NEUROG1, RUNX3 and SOCS1 [Cohen et al. 2016].

Chromosomal instability arises in about 60% of CRCs, and this is a phenomenon that is related to the presence of abnormal chromosome numbers or complements [Pino and Chung, 2010]. MSI is caused by alterations to DNA MMR mechanisms, and it consists of variations in the numbers of repetitive units in each microsatellite sequence. MSI is a characteristic of patients with Lynch syndrome, although it is also associated with about 15–22% of sporadic CRCs [Corso et al. 2013]. Inactivation of the MMR genes (i.e. MLH1, MSH2, MSH6, PMS1, PMS2) through deletions, point mutations, or epigenetic silencing [Aaltonen et al. 1993; Ionov et al. 1993; Thibodeau et al. 1993; Vasen et al. 1999] is responsible for MSI, which causes the failure of the DNA MMR system to repair errors that occur during replication. On the basis of the observed degree of instability, it has been possible to distinguish two well defined MSI phenotypes in particular: MSI-high (MSI-H) and MSI-low (MSI-L; or microsatellite stable, MSS) [Boland and Goel. 2010; Colussi et al. 2013].

Tumor cells with an MSI-positive phenotype accumulate mutations consequent to the defective DNA repair mechanism, and this might activate key oncogenes or inactivate tumor suppressor genes. The TGF-β tumor suppressor gene [Markowitz et al. 1995], the TGF-β type II receptor (TGF-βR2), BAX, caspase 5 apoptotic regulators [Duval and Hamelin, 2002; Vousden and Prives, 2009], and the tumor suppressor gene TCF4 (which has been implicated in deregulation of the Wnt/β-catenin/TCF signaling pathway) are the most frequently altered genes in these tumors [Cuilliere-Dartigues et al. 2006].

The mechanism of MMR gene inactivation is different between sporadic and hereditary cancers. The most frequent MMR gene alteration in sporadic CRCs consists of promoter hypermethylation of the hMLH1 gene, instead of point mutations of the hMLH1 or hMSH2 genes, as described for HNPCC [Markowitz and Bertagnolli, 2009]. The MSI sporadic CRCs are frequently associated with the BRAF V600E mutation, which mediates extensive epigenetic silencing in CRC [Fang et al. 2016]. This therefore represents a diagnostic marker to distinguish sporadic from hereditary CRC [Capper et al. 2013].

Several altered molecular signaling pathways are involved in CRC onset, such as the Wnt, RAS/MAPK, PI3K, TGF-β, NF-κB pathways or DNA MMR genes. These alterations are responsible for responsiveness or resistance to antitumor agents (Figure 2), and they can confer individual susceptibility to cancers when they are germlines [Markowitz and Bertagnolli, 2009; Najdi et al. 2011; De Rosa et al. 2015].

OPEN IN VIEWER

Figure 2.

Functional cross-talk between the signaling pathways involved in colorectal cancer.

Activating mutations of the Ras gene and pathway, including alterations of many of its regulators or effectors, probably represent the most frequent genetic alterations in human tumors [Bos, 1989; Rojas and Santos, 2002; Parsons et al. 2005; Mandal et al. 2015; Moura et al. 2015; Ratner and Miller, 2015]. Ras is a multitasking protein that cooperates with several effectors in the regulation of many cellular processes, such as cell proliferation, differentiation, apoptosis, and senescence [Moodie et al. 1993; Warne et al. 1993; Hallberg et al. 1994;]. The Ras genes (i.e. K-, H- and NRas) are localized on chromosomes 12, 11 and 1, respectively, and encode small proteins with GTPase activity that associate with the plasma membrane.

Ras mutations usually activate Ras signaling by enhancing GTP levels, and in this way mediating phosphorylation and activation of the Raf proteins (A, B and C) (Figure 2) [Vojtek et al. 1993; Zhang et al. 1993; Malumbres and Barbacid, 2003]. It is now believed that oncogenic Ras can promote tumor onset and progression [Chin et al. 1999], although multiple Ras-interacting proteins, such as the PI3Ks, appear to be required to sustain tumor growth [Lim and Counter, 2005]. For example, oncogenic Ras and PI3K cooperate to promote the loss of anoikis, so taking part in the progression of an advanced tumor into metastases [Chiarugi and Giannoni, 2008; Simpson et al. 2008].

The PI3K pathway is often dysfunctional in both sporadic and hereditary CRC, and it activates cell growth, inhibits apoptosis, and regulates cellular transformation, adhesion, motility, and survival [Cantley, 2002; Fruman et al. 1998] in response to several extracellular signals, such as growth factors, cytokines, hormones, heat and oxidative stress, hypoxia, and hypoglycemia (Figure 2) [Testa and Bellacosa, 2001; Martini et al. 2014; Vivanco and Sawyers, 2002; Castellano and Downward, 2011].

Several genes and alternative splicing mechanisms are responsible for the synthesis of the PI3K heterodimeric lipid kinases complex, which consists of the catalytic, adaptor and regulatory subunits. This protein family has three main classes (class I, II, and III), with class I as the best characterized and the most clearly known to be associated with human cancers [Yuan and Cantley, 2008; Castellano and Downward, 2011]. The main reported alterations are recurrent mutations in the PIK3CA gene [Engelman et al. 2006; Samuels et al. 2004], such as gene amplifications, deletions, and somatic missense mutations, which probably increase the kinase activity of PIK3CA, to contribute to cellular transformation [Karakas et al. 2006].

PTEN is a negative regulator of the PI3K/AKT pathway (Figure 2), and is a tumor suppressor gene, alterations to which are involved in several sporadic and hereditary CRCs. As with small quantitative alterations of PTEN expression, somatic mutations in the PTEN gene have been shown to be associated with different kinds of cancers [Eng 2003; Manfredi 2010].

The hepatocyte growth factor (HGF) receptor is known as mesenchymal–epithelial transition factor (cMET), and it belongs to the tyrosine kinase receptor super-family. The cMET binds its ligand, HGF, which leads to its dimerization, phosphorylation, and finally, activation (Figure 2). The cMET proto-oncogene, which is encoded by 21 exons, and is localized on chromosome 7 at position q21–q31 [Duh et al. 1997; Liu, 1998]. This activation of the HGF/cMET pathway induces EMT, with the regulation of specific mechanisms, such as E/N-cadherin switching and extracellular matrix degradation processes. These, in turn, promote cell motility, survival, and proliferation. It is now well known that EMT is a physiological cellular mechanism that tumor cells borrow during the metastatic process. In this way, in many human cancers, the HGF/cMET pathway participates in malignant progression and cell scattering, invasion, and protection from apoptosis, and in metastasis formation [Fujita and Sugano, 1997; Samamé Pérez-Vargas et al. 2013; Smith and Bhowmick, 2016].

Several mechanisms are involved in this cMET activation, and many human cancers have been described with point mutations, gene amplification, mRNA overexpression, or ligand autocrine loops [Samamé Pérez-Vargas et al. 2013]. Another way by which the HGF/cMET signaling pathway contributes to tumor progression and metastases is through stimulation of angiogenesis and lymphangiogenesis, thus acting in a synergistic manner with the VEGF/VEGFR pathway (Figure 2). Moreover, the HGF/cMET signaling pathway promotes the growth of endothelial cells and overexpression of pro-angiogenic factors, such as VEGF itself [Ding et al. 2003; Zhang et al. 2003; Matsumura et al. 2013].

The role of angiogenesis in tumor progression and metastases has been well investigated for a long time. The mammalian VEGF family consists of five glycoproteins (i.e. VEGFA, VEGFB, VEGFC, VEGFD, PlGF), among which VEGFA is the best characterized [Dvorak, 2002; Hicklin and Ellis, 2005; Fan et al. 2005]. In-vitro studies have shown that a monoclonal antibody against VEGFR1 blocks tumor cell migration and invasion in CRC cells that overexpress this receptor [Ellis and Hicklin, 2008]. The VEGFR is a tyrosine kinase protein that binds its ligand and activates downstream signaling pathways that promote pro-angiogenic processes (Figure 2) [Kowanetz and Ferrara, 2006]. A role for VEGF was also recently suggested in the escape of tumors from immune responses, through the block of dendritic cell differentiation [Gabrilovich et al. 1996; Ohm et al. 2003; Ellis and Hicklin, 2008]. Indeed, dendritic cells are antigen-presenting cells that have a critical role in presenting cancer antigens to several immune cell types [Steinman et al. 2003; Banchereau and Palucka, 2005; Colonna et al. 2006].

The microenvironment is mainly represented by the immune system and the extracellular matrix, and it also provides tumor heterogeneity [Vermeulen et al. 2010; Herrera et al. 2013; Berdiel-Acer et al. 2014]. In both localized tumors and the whole organism, the interactions between normal and tumor cells contribute to inter-individual genetic heterogeneity. These interactions are the reason for the different responses between individuals with apparently similar tumors, which are due to their different genetic backgrounds.

Colorectal cancer heterogeneity can be observed among antitumor immune surveillance systems, mainly in terms of the major histocompatibility complex, although this heterogeneity is also seen for other important molecules that are dedicated to immunological regulation, such as galectin-3 and programmed death ligand-1 [Bernal et al. 2011]. There is now strong evidence that subcellular populations showing CD3+, CD8+, and CD45RO+ antigens have a central role in antitumor immunological responses. Altered expression of major histocompatibility complex antigens represents an early event in CRC tumorigenesis. This induces the so-called phase of ‘escape’, in which clones of tumor cells can evade the immunological control and initiate metastatic spreading [Garrido et al. 1993, 1997]. Interestingly, the presence of peritumoral inflammatory infiltrate is a characteristic of MSI tumors that correlates with a good prognosis. This confirms that tumor immunological characteristics and responses in CRC are related to the genetic background of the tumor [Sinicrope and Sargent, 2012].

As described above, during EMT, the cancer cells can detach from the primary cancer, migrate into the surrounding tissues, and start the metastatic development that is characterized by intravasation, dissemination, extravasation, and colonization. Thus, once in the vessels, these cancer cells can circulate in the blood. CRCs show great heterogeneity not only in the primary tumor, but also in these circulating cancer cells. Indeed, they show different differentiation grades and epithelial, mesenchymal, or stem cell biomarker expression [Yu et al. 2013]. Investigations into the molecular biomarkers of these circulating cancer cells represent an interesting field in CRC clinical research. The cancer cells that show mesenchymal and stem cell-like phenotypes are more able to undergo extravasation and colonization of distant organs, compared with those with an epithelial phenotype [Tsai and Yang, 2013; Kalikaki et al. 2014].

As described above, CRCs are characterized by this great genotypic and phenotypic heterogeneity that includes the tumor microenvironment and the interactions between healthy and cancer cells. All of these traits confer unique peculiarities to each tumor, and thus each tumor can be considered as a single disease. Here, well conducted molecular and clinical characterization is essential, with the view to implement precision oncology, and thus personalized care.

Prognostic and predictive biomarkers in colorectal cancer

Conclusion

As described so far, several alterations at the molecular level favor CRC onset and progression, and the formation of metastases. Thus, molecular characterization of cancer-associated mutations can provide valuable information on disease prognosis and patient response to therapy. The advent of NGS technology has allowed better, more rapid, and also cheaper, molecular characterization of cancers, and clinically-relevant biomarkers can be measured. In summary, germline mutational screening of genes responsible for hereditary and familial CRC represent essential molecular tests for cancer predisposition, while several somatic predictive/ prognostic molecular biomarkers are available today. As described above, KRAS mutations are the only molecular marker actually validated and accepted in clinical practice as a predictive biomarker in response to EGFR inhibition. However, the ‘Quadruple index’, which associates EGFR monoclonal antibody responses with changes in one or more of four genes, as KRAS, BRAF, PIK3CA and PTEN, might provide further information. The MSI status has a role as a prognostic biomarker, although its role as a predictive biomarker for response to therapy remains to be clarified. In our opinion, the TNM-I system coupled with tumor budding might represent the best pathological prognostic biomarkers, which need to be validated for clinical practice. Furthermore, stemness and EMT molecular biomarkers might not only improve cancer screening and early detection, but might also help to define the best postsurgery follow up, not only for patients with advanced and mCRC, but also for patients with early stage disease, such as node-negative patients. Only stem cells have the potential for unlimited proliferation, multi-lineage differentiation, and colonization of new sites, and thus these represent the most likely precursors for invasive and mCRC, and therefore these CSCs require much further attention.

As we have previously reported, a multidisciplinary approach that includes genetics, molecular biology, surgery, and clinical oncology is the only way to achieve the best results for patient outcomes [De Rosa et al. 2015]. On this issue, future studies in CRC need to be concentrated on the standardization of procedures to better determine the genetic profile of each cancer and to define a panel of predictive/ prognostic biomarkers that will allow surgeons and clinical oncologist to choose the best treatment strategy for each, and every single patient.