Introduction
The face of ‘cutting edge’ medicine is rapidly changing. In a recent paper in which they reflect on the changes taking place, Stephen Friend and Leroy Hood [1] refer to this as P4 medicine: predictive, personalized, preventive and participatory. Much of this has been brought about by the molecular genetic revolution. It is now possible to sequence a whole genome in a matter of days and at an affordable cost. In the near future we might all have our personal genomic information with us on a smart card, containing the relevant elements of our personal medical history. We will focus here on colorectal cancer (CRC), one of the most frequently occurring cancers, globally responsible for a high number of cancer-associated deaths annually [2]. Colorectal cancer is one of the most intensively studied cancer types, partly because of its high prevalence but also because of the existence of its precursor lesions, tubular or villous adenomas and more recently (sessile) serrated adenomas, which can be detected endoscopically and removed. Theoretically, removal of these adenomas would prevent most cases of CRC from developing. Characteristic morphological steps in the evolution of these precursor lesions have been elucidated at a molecular level, and the adenoma-carcinoma sequence, as this has become known, is one of the classical examples of stepwise progression of cancer. Gaining this knowledge has been facilitated by the occurrence of a variety of forms of familial intestinal cancer, the molecular genetic background of which has been largely clarified [3]. Apart from early detection of familial forms of CRC and its use in genetic counseling, until recently detailed molecular knowledge has had little impact on the clinical management of CRC. Classical clinico-pathological parameters have remained essential factors in determining how a CRC patient will be treated. It is the purpose of this paper to review how advancing knowledge of the molecular events involved in the development of CRC and its precursor lesions up to the metastatic stage have influenced our approach to (early) detection and treatment of this globally widespread disease, which still kills half of those who develop it [2].
Pathways in the development of colorectal cancer
Molecular studies of CRC have elucidated that several signaling pathways are involved in its development. We now have a fairly detailed insight into the main (epi)genetic events involved in colorectal carcinogenesis. What has become clear is that CRC is not a single disease [3]. The morphological heterogeneity, in terms of site, grade and type of the tumor, has been expanded by several added layers of molecular complexity [4-6]. We now realize that CRC is a very heterogeneous disease, in terms of clinical presentation, likelihood of cure, pattern of extension and response to treatment, to name but a few determinants of this heterogeneity. A structured approach towards understanding at least parts of this heterogeneity follows the molecular pathways involved in its genesis. Three main pathway concepts have now been developed [7-9], and a fourth has been recently added [10].
The chromosomal instability pathway
The first elucidated is the classical pathway, of which chromosomal instability (CIN) is a hallmark. These carcinomas are usually highly aneuploid and show striking variation in gene copy number. The concept of this pathway was developed by Vogelstein in a landmark paper [7], and it is involved in roughly 80% of colorectal carcinomas. The events in this pathway are illustrated in Fig. 1. Elucidation of the pathway was facilitated through the discovery of the adenomatous polyposis coli (APC) gene, the gene responsible for familial adenomatous polyposis (FAP) [11, 12]. APC mutations activate the Wnt signaling pathway, which plays a central role in the development of CRC, as in about 90% of all CRC cases gene abnormalities occur which activate this pathway [13]. Most of these are loss of function mutations of APC, but rare -catenin mutations have been reported. The second allele of APC is subsequently silenced through mutation, allelic loss or gene promoter methylation. Loss of function of the APC protein interferes with phosphorylation of -catenin, which is then no longer ubiquitinated and fails to be eliminated in the proteasome. When -catenin accumulates, it is translocated to the nucleus, where it assumes a transcription factor function in interaction with LEF/TCF and upregulates the expression of genes that promote cell growth (Fig. 2). Wnt pathway activation is already present in adenomatous polyps, precursor lesions of CRC, and hence is considered as an early event in CRC carcinogenesis.
That colon adenomas are precursor lesions of carcinomas was known, long before molecular knowledge started to elucidate the mechanisms involved. Over time, the progression of an adenomatous polyp into a malignant carcinoma became known as the adenoma-carcinoma sequence. With the elucidation of molecular mechanisms involved, Vogelstein [7] supplemented this concept with molecular data, which with the equally well established stepwise progression concept resulted in the molecular model which has become almost a standard approach to address molecular carcinogenesis (Fig. 1).
FAP patients with a germline mutation of APC develop hundreds to thousands of adenomatous polyps already at a very young age and have a 100% lifetime risk of the development of CRC. Development of neoplastic lesions is not limited to the colorectum: neoplasms occur also in the upper intestinal tract (stomach, duodenum) and in connective tissue (desmoid tumors). There is a striking genotype-phenotype relationship regarding the clinical characteristics of FAP patients. Mutations before codon 157, after codon 1595, and in the alternatively spliced region of exon 9 are associated with a much lower adenoma frequency (attenuated FAP, for a review see 14) and also with the occurrence of lesions in the upper gastrointestinal tract (duodenum and stomach). In contrast, mutations between codons 1250 and 1464 are associated with severe polyposis. Mutations after codon 1444 are associated with occurrence of desmoid tumors [14]. Also in sporadic CRC, which usually develop at advanced age (mostly from the seventh decade on), APC mutations occur but these are somatic, not germline [15].
Additional molecular events are activating mutations of KRAS and BRAF in the MAPK pathway. Mutations of KRAS are found in around 45% of CRC and constitutively activate the MAPK signaling pathway (Fig. 3). BRAF, downstream of KRAS in the MAPK pathway, is mutated in less than 10% of the cases. Mutations in the MAPK pathway also occur in NRAS and in phosphoinositide-3-kinase (PIK3) [16]. We will come back to this issue in the context of molecular predictors of response to targeted treatment in CRC.
In the classical pathway, loss of function mutations of the TP53 gene occur in about 70% of the cases with accumulation of the mutated protein in the nucleus (Fig. 4), but as a rule in the progression from high-grade adenoma to carcinoma, when also telomerase is activated [17], which confers an unlimited lifespan on the transformed cells. Activation of the TGF- pathway, which is reflected in the loss of its downstream effector SMAD4, is probably involved in metastatic progression [18].
The classical pathway is associated with FAP and germline mutations in the APC gene, but not all patients with polyposis have an APC mutation. Detailed analysis of the genome of such patients has led to the discovery of the involvement of the MUTYH gene, which is involved in base excision repair following oxidative damage to DNA [19]. MUTYH polyposis (or MAP) to a large extent follows the same sequence of events as FAP patients, but MAP patients often have fewer polyps, carcinomas appear between the fourth and seventh decade, and the lifetime risk for the development of CRC is lower than for FAP patients [20].
The microsatellite instability (MIN) pathway
This pathway is responsible for the development of hypermutating carcinomas, which in contrast to the CIN carcinomas usually show little aneuploidy. The prototype of these pathways was discovered in the early 1990s while searching for the molecular genetic origins of hereditary non-polyposis colon cancer (HNPCC), which is now known as Lynch syndrome [21]. A landmark discovery was that Lynch syndrome is caused by a mutation in one of the genes encoding the proteins involved in mismatch repair (MMR) [22].
During DNA replication errors occur, such as single base mismatches or deletions and short insertions, which are corrected by MMR. A protein complex is formed by the proteins involved in MMR which binds the mismatch and uses the information from the (correct) complementary strand to excise the error and repair it. When MMR does not function, the cells accumulate errors, which occur also in microsatellite sequences. These are repeating sequences of DNA, mostly 1-6 base pairs in length, which occur throughout the genome. MMR deficiency generates novel microsatellite sequences, which can be detected by a PCR-based assay. In this assay a panel of microsatellite marker loci, including BAT25, BAT26, D2S123, D5346, and D17S250, is used to detect microsatellite instability (MSI) [23]. When two or more markers are unstable, it is called MSI-H(igh); when only one marker is instable, it is called MSI-L(ow); and when all markers are stable, it is called microsatellite stable (MSS) (Fig. 5). As mismatch repair proteins are easily detectable in routinely processed tissue sections, immunohistochemical detection of loss of expression of one of the relevant proteins (MLH1, MSH2, MSH6 and PMS2) has become mainstream in the detection of mismatch repair and a screening tool for the detection of Lynch syndrome [24] (Fig. 5). In pathway terms, this MSI-H is equivalent to the microsatellite instability (MIN) pathway. Lynch syndrome is responsible for about 3% of CRC, but MSI occurs in around 12% of sporadic CRC cases. This is due to promoter methylation of MLH1, which shuts down MLH1 transcription, resulting in deficient MMR and MSI-H [25].
MSI carcinomas display a relatively high frequency of BRAF gene mutations. There are other genes preferentially mutated, often characterized by mononucleotide tracts in their coding regions, as is the case for the TGF- receptor II gene and the pro-apoptotic gene BAX, which are as a consequence frequently inactivated in MMR-deficient cancer. The MS status has gained quite a bit of clinical interest for sporadic CRC lately as mismatch repair deficient carcinomas have a better prognosis than mismatch repair competent carcinomas. Microsatellite instability status is gradually entering into clinical decision making [26].
In the MIN pathway, the development of neoplastic lesions is quite similar to that in the CIN pathway: an adenoma-carcinoma sequence. However, cancers that arose through this pathway behave differently from those in the CIN pathway: they have a better prognosis, respond differently to standard chemotherapy, have fairly characteristic morphology as they are situated in the right colon, are of mucinous or medullary histology (Fig. 6), and display a characteristic lymphocytic infiltrate [27].
It had been noted in the The Cancer Genome Atlas (TCGA) whole genome sequencing effort for CRC that a subgroup of hypermutating carcinomas are not microsatellite instable. In addition, families were identified with oligopolyposis and microsatellite stable CRC at young age but without APC or MYH mutations. This stimulated the search for additional mechanisms responsible for a hypermutating state. Recently it was found that germline mutations in the proofreading domain of two DNA polymerases (POLE and POLD1) are associated with this syndrome [10]. Lack of correction of mistakes in DNA replication apparently generates mutations, with a mutator phenotype as a result [28]. This syndrome has been called polymerase proofreading-associated polyposis (PPAP).
The CpG island hypermethylation phenotype pathway
Hypermethylation of gene promoter associated CpG islands, involved in the regulation of transcription, is a frequently encountered mechanism responsible for silencing of tumor suppressor genes. If this phenomenon occurs genome-wide, it is called CpG island hypermethylation phenotype (CIMP) [29]. What causes the CIMP phenotype and what exactly its consequences are remains elusive.
Around the turn of the century, it was noted that flat non-adenomatous mucosal lesions, notably in the right colon, often have a CIMP. These lesions occur as a result of abnormal retention of surface epithelium due to hyperproliferation and inhibition of apoptotic cell loss. They are composed of glands with a saw-tooth like appearance (hence the name serrated) and were long considered as largely innocuous hyperplastic polyps. More recently it was shown that a subtype of them with particular morphology has significant malignant potential, and hence these are called sessile serrated adenomas or polyps (SSA/P) [30]. They resemble (benign) hyperplastic polyps but show a more irregular crypt architecture (Fig. 7) and occasionally features of dysplasia. SSA/P have been associated with increased risk for the development of CRC. The extent of the risk has not yet been adequately established, but current evidence suggests that it is lower than the risk of progression of traditional adenomas [31]. Carcinomas with CIMP were noted to be associated with serrated precursor lesions, and in addition CIMP carcinomas often harbor BRAF mutations. Putting CIMP and serrated morphology together led to the development of the CIMP pathway concept [32].
The molecular events in the pathway include in addition to CIMP mutation of the BRAF gene, characteristically the V600E mutation, methylation of the promoter of a variety of genes (as a result of CIMP), and subsequent methylation of the MLH1 gene promoter with silencing of the expression of MLH1 as a result. This is then responsible for MMR insufficiency, MSI, and a hypermutating state.
It is important to emphasize here that these pathway notions constitute a model framework which is useful in further defining molecular events in colorectal cancer in association with clinical and pathological parameters. In reality, a variety of subgroups of colorectal cancer exist with molecular characteristics which overlap to a significant degree [30]. Recently, several approaches to the development of molecular classifications of CRC have been published. These not surprisingly show significant similarities between each other and they also align with some of the pathway characteristics [4–6]. These molecular classifications add a layer of complexity to the heterogeneity of CRC. Currently, efforts are underway to develop a consensus molecular classification in the context of open source collaborative genomic data analysis of which the first results have already been published [33]. It remains to be seen how exactly these molecular classifications will impact on clinical management of CRC, which is still largely determined by classical TNM parameters, even though the MS status recently has gained importance as a molecular parameter to be taken into account in the choice of adjuvant treatment of stage II/III CRC and potentially also the choice of treatment of metastatic CRC. This brings us to the question of whether, and if so how, these pathway concepts have impacted on the clinical management of CRC.
Molecular solutions to problems in the management of colorectal cancer?
Understanding disease is what pathology ultimately strives for, but this understanding should serve patients through the development of applications from which they profit. The issue is therefore which burning questions in the present modalities of care for CRC patients might profit from a molecular solution. In this context the following questions are relevant:
• Can molecular pathology provide more efficient approaches towards population screening for CRC?
• Can molecular tests predict which patients are at high risk for development of recurrent disease after primary surgery?
• Can molecular tests predict which individuals of the cohort of patients at high risk for recurrent disease most likely will respond to currently used adjuvant (chemo) therapy?
• Can molecular tests predict which patients most likely will respond to the available modalities of targeted treatment?
Molecular screening for colorectal cancer
Colorectal cancer screening programs have risen to global prominence, given the high prevalence of this disease and the possibility to identify and adequately treat precursor lesions by colonoscopy. The standard approach for these screening programs is fecal occult blood testing (FOBT), in order to identify a more limited population qualifying for colonoscopy in case of a positive FOBT. Fecal occult blood testing is relatively sensitive but lacks specificity [34]. A variety of molecular approaches are under investigation to develop more specific tests. Most focus on the detection of tumor DNA in fecal matter, looking for tumor-specific gene alterations such as hypermethylation or specific mutations. In a laboratory setting such tests have shown promising results, but as yet sensitive and specific molecular tests to detect cancer are not available [35].
Molecular prognostic tests for colorectal cancer
A key problem in the clinical management of most cancers is the question of which patients will present recurrent disease after initial therapy, which for most cancers is surgical removal of the primary tumor, occasionally after neoadjuvant (chemo- and/or radio-)therapy. The issue is clear: those patients with a high likelihood of recurrent disease, which for now means stage III and high-risk stage II cases, might be treated with adjuvant chemotherapy. Those with a low likelihood of recurrent disease, stage I and low-risk stage II
patients, should be spared this therapy, which is associated with significant side effects. Presently used criteria used for patient stratification lack accuracy, as even among stage I cancers a small proportion will relapse and about 30% of stage III patients will not relapse [36]. Several prognostic molecular signatures have been developed, which potentially would allow stratification of patients according to the risk of recurrence [37-39]. A recent study has independently assessed the performance of these signatures in an independent patient population and concluded that all were prognostically significant [40]. Strikingly, however, they did not use similar gene sets, nor did they identify the same patients as high risk. A combination of these signatures, not surprisingly, performed better than any single signature. These data indicate that even though finding effective (molecular) approaches towards identification of high-risk CRC patients remains urgent, as yet the efficacy of attempts to attain this goal remains questionable.
Molecular tests predictive for response to chemotherapy
Of the CRC patients who qualify for adjuvant chemotherapy (the ‘high-risk’ – based on clinical and histopathological criteria – stage II and the stage III patients), only about 50% respond [41]. Those that do not respond are exposed to the usual side-effects of chemotherapy without having any benefit. It would therefore be valuable to have (clinical, molecular) parameters predicting response to this regimen, which as a rule consists of a combination of 5-fluorouracil, leucovorin and oxaliplatin (known as FOLFOX). Some attempts have been made to develop molecular tests to this effect, but as yet none are used in daily practice due to lack of sensitivity or specificity [42, 43]. The search therefore remains on for more effective approaches towards adjuvant treatment as well as for tools to predict treatment response.
Molecular markers predictive of a response to targeted therapy in colorectal cancer
As stated before, the standard therapy for advanced CRC is based on the combination of 5-fluorouracil/leucovorin/oxaliplatin (FOLFOX). In the last decade, evidence was obtained that for a subpopulation of patients a therapeutic approach addressing the epidermal growth factor receptor (EGFR) might be advantageous. Monoclonal antibodies were developed against EGFR with some efficacy in CRC. In prospective clinical trials, initial observations were validated, initially showing a response to cetuximab [44] and subsequently also panitumumab [45] but only in patients with a KRAS wild type tumor. This finding rapidly promoted KRAS mutation testing in CRC to one of the most frequently performed tests in molecular pathology laboratories. An important limitation of the KRAS mutation status is, however, that it identifies (KRAS-mutated) patients who will not respond to anti-EGFR therapy, which eliminates about 40% of the non-responders, but of the KRAS-wild type patients only about 40% respond. The capacity to effectively predict who will respond rather than those patients who will not would be an improvement. Assessing the mutation status of other genes in the MAPK pathway has been investigated [46], and an approach combining testing of the mutation status of several genes involved in the MAPK pathway effectively increases the predictive value. The accuracy of the presently available test approaches is, however, still far from ideal, and the search is on for molecular tests that more accurately predict targeted therapy response [47].
An important issue in this context is therapy resistance. The beneficial effect of anti-EGFR is as a rule of short duration because of the development of resistance [48]. Ongoing studies explore the mechanisms responsible for this treatment failure [49]. Others search for new molecular targets to treat advanced colorectal cancer [50].
Concluding remarks
We are not at the end of what has been a fascinating adventure in terms of discovering the details of colorectal carcinogenesis. We have come from classical histopathology (including key gross pathology parameters) through a continuously more refined TNM classification approach to molecular parameters to stratify patients for therapy, befitting the specific characteristics of their cancer. We might not understand colorectal carcinogenesis completely, but our knowledge is much better than it was two decades ago. We are in the middle of the development of parameters predicting response to available targeted therapies, of the development of new therapies targeting elements of the pathways involved in colorectal carcinogenesis, and of molecular tests that will predict the response to them. Global attempts are being made through TCGA project to gain more insight into the molecular biology of colorectal (and many other) cancers. Pathology plays a key role in this development, as much of the knowledge that has been gained has been (and still is) based on molecular studies on tissue samples that pathologists use to make a diagnosis. Pathology and pathologists should remain in that central position as a partner in a multidisciplinary effort, along with molecular (cancer) biologists, bioinformaticians and (molecular) oncologists. Eventually, this will improve early detection and provide more accurate prediction of (adjuvant) treatment need, more efficient drugs, and accurate predictors of drug response, from which our patients will profit.
References
Primarily timely reviews have been chosen, rather than original papers to facilitate further exploration of the relevant literature.
1. Hood L, Friend SH. Predictive, personalized, preventive, participatory (P4) cancer medicine Nat Rev Clin Oncol 2011; 8: 184-187.
2. Bosman FT. Colorectal cancer. In: World Cancer Report 2014 (Ed. Stewart BW and Wild CP). IARC Press, Lyon 2014; 32-402.
3. Fearon ER Molecular genetics of colorectal cancer. Annu Rev Pathol Mech Dis 2011; 6: 479-507.
4. Sadanandam A, Lyssiotis CA, Homicsko K, et al. A colorectal cancer classification system that associates cellular phenotype and responses to therapy. Nat Med 2013; 19: 619-625.
5. De Sousa E Melo F, Wang X, Jansen M, et al. Poor-prognosis colon cancer is defined by a molecularly distinct subtype and develops from serrated precursor lesions. Nat Med 2013; 19: 614-618.
6. Budinska E, Popovici V, Tejpar S, et al. Gene expression patterns unveil a new level of molecular heterogeneity in colorectal cancer. J Pathol 2013; 231: 63-76.
7. Fearon ER, Vogelstein B. A genetic model for colorectal carcinogenesis. Cell 1996; 87: 159-170.
8. Hemminki A, Peltomäki P, Mecklin JP, et al. Loss of the wild type MLH1 gene is a feature of hereditary nonpolyposis colorectal cancer. Nat Genet 1994; 8: 405-410.
9. Jass JR. Serrated adenoma of the colorectum and the DNA-methylator phenotype. Nat Clin Pract Oncol 2005; 2: 398-405.
10. Briggs S, Tomlinson I. Germline and somatic polymerase and mutations define a new class of hypermutated colorectal and endometrial cancers. J Pathol 2013; 230: 148-153.
11. Powell SM, Zilz N, Beazer-Barclay Y, et al. APC mutations occur early during colorectal tumorigenesis. Nature 1992; 359: 235-237.
12. Cottrell S, Bicknell D, Kaklamanis L, Bodmer WF. Molecular analysis of APC mutations in familial adenomatous polyposis and spoadic colon carcinomas. Lancet 1992; 340: 626-630.
13. Giles RH, van Es JH, Clevers H. Caught up in a Wnt storm: Wnt signaling in cancer. Biochim Biophys Acta 2003; 1653: 1-24.
14. Nieuwenhuis MH, Vasen HF. Correlations between mutation site in APC and phenotype of familial adenomatous polyposis (FAP): a review of the literature. Crit Rev Oncol Hematol 2007; 61: 153-161.
15. Miyoshi Y, Nagase H, Ando H, et al. Somatic mutations of the APC gene in colorectal tumors: mutation cluster region in the APC gene. Hum Mol Genet 1992; 1: 229-233.
16. Tian S, Simon I, Moreno V, et al. A combined oncogenic pathway signature of BRAF, KRAS and PI3KCA mutation improves colorectal cancer classification and cetuximab treatment prediction. Gut 2013; 62: 540-549.
17. Yan P, Saraga EP, Bouzourene H, et al. Telomerase activation in colorectal carcinogenesis. J Pathol 1999; 189: 207-212.
18. Fleming NI, Jorissen RN, Mouradov D, et al. SMAD2, SMAD3 and SMAD4 mutations in colorectal cancer. Crit Rev Oncol Hematol 2011; 79: 1-16.
19. Nielsen M, Morreau H, Vasen HF, Hes FJ. MUTYH-associated polyposis (MAP). Cancer Res 2013; 73: 725-735.
20. Sampson JR, Jones N. MUTYH-associated polyposis. Best Pract Res Clin Gastroenterol 2009; 23: 209-218.
21. Lynch HT, Lanspa SJ, Boman BM, et al. Hereditary nonpolyposis colorectal cancer – Lynch syndromes I and II. Gastroenterol Clin North Am 1988; 17: 679-712.
22. Nyström-Lahti M, Parsons R, Sistonen P, et al. Mismatch repair genes on chromosomes 2p and 3p account for a major share of hereditary nonpolyposis colorectal cancer families evaluable by linkage. Am J Hum Genet 1994; 55: 659-665.
23. Mead LJ, Jenkins MA, Young J, et al. Microsatellite instability markers for identifying early-onset colorectal cancers caused by germ-line mutations in DNA mismatch repair genes. Clin Cancer Res. 2007; 13: 2865-2869.
24. Shia J, Tang LH, Vakiani E, et al. Immunohistochemistry as first-line screening for detecting colorectal cancer patients at risk for hereditary nonpolyposis colorectal cancer syndrome: a 2-antibody panel may be as predictive as a 4-antibody panel. Am J Surg Pathol 2009; 33: 1639-1645.
25. Cunningham JM, Christensen ER, Tester DJ, et al. Hypermethylation of the hMLH1 promoter in colon cancer with microsatellite instability. Cancer Res 1998; 58: 3455-3460.
26. Pino MS, Chung DC. Microsatellite instability in the management of colorectal cancer. Expert Rev Gastroenterol Hepatol 2011; 5: 385-399.
27. Tougeron D, Maby P, Elie N, et al. Regulatory T lymphocytes are associated with less aggressive histologic features in microsatellite-unstable colorectal cancers. PLoS One 2013; 8: e61001.
28. Heitzer E, Tomlinson I. Replicative DNA polymerase mutations in cancer. Curr Opin Genet Dev 2014; 24: 107-113.
29. Teodoridis JM, Hardie C, Brown R. CpG island methylator phenotype (CIMP) in cancer: causes and implications. Cancer Lett 2008; 268: 177-186.
30. Snover DC. Update on the serrated pathway to colorectal carcinoma. Hum Pathol 2011; 42: 1-10.
31. Bettington M, Walker N, Clouston A, et al. The serrated pathway to colorectal carcinoma: current concepts and challenges. Histopathology 2013; 62: 367-386.
32. Jass JR, Baker K, Zlobec I, et al. Advanced colorectal polyps with the molecular and morphological features of serrated polyps and adenomas: concept of a ‚fusion’ pathway to colorectal cancer. Histopathology 2006; 49: 121-131.
33. Sadanandam A, Wang X, de Sousa E Melo F, et al. Reconciliation of classification systems defining molecular subtypes of colorectal cancer: interrelationships and clinical implications. Cell Cycle 2014; 13: 353-357.
34. Pox CP. Controversies in colorectal cancer screening. Digestion 2014; 89: 274-281.
35. Imperiale TF, Ransohoff DF, Itzkowitz SH, et al. Multitarget stool DNA testing for colorectal-cancer screening. N Engl
J Med 2014; 370: 1287-1297.
36. Dotan E, Cohen SJ. Challenges in the management of stage II colon cancer. Semin Oncol 2011; 38: 511-520.
37. Omberg L, Ellrott K, Yuan Y, et al. Enabling transparent and collaborative computational analysis of 12 tumor types within The Cancer Genome Atlas. Nat Genet 2013; 45: 1121-1126.
38. Maak M, Simon I, Nitsche U, et al. Independent validation of a prognostic genomic signature (ColoPrint) for patients with stage II colon cancer. Ann Surg 2013; 257: 1053-1058.
39. Reimers MS, Kuppen PJ, Lee M, et al. Validation of the 12-gene colon cancer recurrence score as a predictor of recurrence risk in stage II and III rectal cancer patients. J Natl Cancer Inst 2014; 106 pii: dju269. doi: 10.1093/jnci/dju269.
40. Di Narzo AF, Tejpar S, Rossi S, et al. Test of four colon cancer risk-scores in formalin fixed paraffin embedded microarray gene expression data. J Natl Cancer Inst 2014; 106 pii: dju247. doi: 10.1093/jnci/dju247.
41. Van Loon K, Venook AP. Adjuvant treatment of colon cancer: what is next? Curr Opin Oncol 2011; 23: 403-409.
42. Yuan Y, Tan CW, Shen H, et al. Identification of the biomarkers for the prediction of efficacy in first-line chemotherapy of metastatic colorectal cancer patients using SELDI-TOF-MS and artificial neural networks. Hepatogastroenterology 2012; 59: 2461-2465.
43. Jensen NF, Smith DH, Nygård SB, et al. Predictive biomarkers with potential of converting conventional chemotherapy to targeted therapy in patients with metastatic colorectal cancer. Scand J Gastroenterol 2012; 47: 340-355.
44. Lièvre A, Bachet JB, Le Corre D, et al. KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Res 2006; 66: 3992-3995.
45. Peeters M, Douillard JY, Van Cutsem E, et al. Mutant KRAS codon 12 and 13 alleles in patients with metastatic colorectal cancer: assessment as prognostic and predictive biomarkers of response to panitumumab. J Clin Oncol 2013; 31: 759-765.
46. De Roock W, Claes B, Bernasconi D, et al. Effects of KRAS, BRAF, NRAS, and PIK3CA mutations on the efficacy of cetuximab plus chemotherapy in chemotherapy-refractory metastatic colorectal cancer: a retrospective consortium analysis. Lancet Oncol 2010; 11: 753-762.
47. Sartore-Bianchi A, Bencardino K, Di Nicolantonio F, et al. Integrated molecular dissection of the epidermal growth factor receptor (EGFR) [corrected] oncogenic pathway to predict response to EGFR-targeted monoclonal antibodies in metastatic colorectal cancer. Target Oncol 2010; 5: 19-28.
48. Van Emburgh BO, Sartore-Bianchi A, Di Nicolantonio F, et al. Acquired resistance to EGFR-targeted therapies in colorectal cancer. Mol Oncol 2014; 8: 1084-1094.
49. Bardelli A, Siena S. Molecular mechanisms of resistance to cetuximab and panitumumab in colorectal cancer. J Clin Oncol 2010; 28: 1254-1261.
50. Chong CR, Jänne PA. The quest to overcome resistance to EGFR-targeted therapies in cancer. Nat Med 2013; 19: 1389-1400.
Address for correspondence
Fred T. Bosman
Institute of Pathology, University of Lausanne Medical Center
1011 Lausanne
Switzerland
e-mail: Fred.bosman@chuv.ch
