colorectal cancer molecular biology moves into clinical

doi: 10.1136/gut.2009.206250 published online October 4, 2010Gut

Colin C Pritchard and William M Grady clinical practiceColorectal cancer molecular biology moves into

http://gut.bmj.com/content/early/2010/10/04/gut.2009.206250.full.htmlUpdated information and services can be found at:

These include:

References http://gut.bmj.com/content/early/2010/10/04/gut.2009.206250.full.html#ref-list-1

This article cites 111 articles, 54 of which can be accessed free at:

P<P Published online October 4, 2010 in advance of the print journal.

serviceEmail alerting

box at the top right corner of the online article.Receive free email alerts when new articles cite this article. Sign up in the

Topic collections

(32 articles)GUT Recent advances in basic science � Articles on similar topics can be found in the following collections

Notes

articles must include the digital object identifier (DOIs) and date of initial publication. priority; they are indexed by PubMed from initial publication. Citations to Advance online prior to final publication). Advance online articles are citable and establish publicationyet appeared in the paper journal (edited, typeset versions may be posted when available Advance online articles have been peer reviewed and accepted for publication but have not

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://journals.bmj.com/cgi/epTo subscribe to BMJ go to:

group.bmj.com on October 10, 2010 - Published by gut.bmj.comDownloaded from

http://gut.bmj.com/content/early/2010/10/04/gut.2009.206250.full.html

http://gut.bmj.com/content/early/2010/10/04/gut.2009.206250.full.html#ref-list-1

http://gut.bmj.com/cgi/collection/gut_recent_advances_in_basic_science

http://group.bmj.com/group/rights-licensing/permissions

http://journals.bmj.com/cgi/reprintform

http://journals.bmj.com/cgi/ep

http://gut.bmj.com/

http://group.bmj.com/

Colorectal cancer molecular biologymoves into clinical practiceColin C Pritchard,1 William M Grady2,3

ABSTRACTThe promise of personalised medicine is now a clinicalreality, with colorectal cancer genetics at the forefront ofthis next major advance in clinical medicine. This is nomore evident than in the recent advances in testing ofcolorectal cancers for specific molecular alterations inorder to guide treatment with the monoclonal antibodytherapies cetuximab and panitumumab, which target theepidermal growth factor receptor. In this review, geneticmechanisms of colorectal cancer and how thesealterations relate to emerging biomarkers for earlydetection and risk stratification (diagnostic markers),prognosis (prognostic markers) and the prediction oftreatment responses (predictive markers) are examined.

INTRODUCTIONThe promise of personalised medicine is nowa clinical reality, with colorectal cancer genetics atthe forefront of this next major advance in clinicalmedicine. This is no more evident than in thetesting of colorectal cancers for specific molecularalterations in order to guide treatment with themonoclonal antibody therapies cetuximab andpanitumumab, which target the epidermal growthfactor receptor (EGFR).1e3 Indeed, the discoverythat acquired KRAS mutations are a robustpredictive marker of resistance to cetuximab andpanitumumab4 5 has led to clinically validated andcost-effective testing strategies to direct these drugsto appropriate patients. This discovery resultedfrom a detailed understanding of colorectal cancergenetics, including the role of KRAS mutations incolorectal carcinogenesis, as well as knowledge ofthe EGFR signalling pathways.6 The success ofKRAS mutation testing in predicting treatmentresponse is just the beginning of the use of geneticmarkers for directing the care of patients withcolorectal cancer. Many other genetic markers incolorectal cancer show promise for their use intreatment selection, prognosis and early cancerdetection. In this context, knowledge of theunderlying genetic mechanisms of colorectaltumorigenesis and the potential of specific geneticlesions for clinical decision making is expected tobecome part of the working knowledge of careproviders managing colon cancer patients.However, despite the promising advances in themolecular pathology of colorectal cancer that arehighlighted in this review, it is important toemphasise that clinicopathological staging oftumour tissue is still the cornerstone of prognosti-cation and treatment selection. The moderntumourenodeemetastasis (TNM) classification

system is recommended, although the originalDukes staging system is still used by some clini-cians and is taught to pathologists in training.7 Thepathological features with greatest prognosticpower are depth of tumour invasion, burden oflymphovascular invasion (estimated by the numberof lymph nodes infiltrated by cancer) and presenceof distant metastases. Efforts to correlate geneticalterations with histological features have hadlimited success, although microsatellite instabilityis a molecular feature that shows modest correla-tion with certain histological features such ascribriform architecture and medullary histology.8

Thus, molecular testing is usually required foraccurate assessment of specific gene mutations orgenomic instability that provide prognostic andpredictive information beyond clinicopathologicalfeatures.

In this review, we examine genetic mechanisms ofcolorectal cancer and how these alterations relate toemerging biomarkers for early detection and riskstratification (diagnostic markers), prognosis (prog-nostic markers) and the prediction of treatmentresponses (predictive markers) (table 1, box 1). Thegenetic features of colorectal cancer that arecurrently most clinically useful will be emphasisedin this review, and a detailed description of themolecular genetics and molecular biology of thegermane genetic and epigenetic alterations will beprovided. We will conclude by reviewing thepotential role of genetic markers in the selection oftargeted colorectal cancer treatments that are inpreclinical development or in phase I and II trials.

MOLECULAR MECHANISMS OF COLORECTALCARCINOGENESISThe adenoma/carcinoma progression sequenceColorectal cancer arises as the result of the accu-mulation of acquired genetic and epigeneticchanges that transform normal glandular epithelialcells into invasive adenocarcinomas. Steps thattransform normal epithelium into benign neoplasia(adenoma), followed by invasive carcinoma andeventually metastatic cancer are described in theclassic tumour progression model proposed byFearon and Vogelstein (figure 1).6 Since this modelwas originally proposed, our understanding of themolecular pathogenesis has advanced considerablyand led to numerous revisions of the Vogelstein andFearon model. For instance, the original modelproposed that only tubular and tubulovillousadenomas had the potential to progress to invasiveadenocarcinoma. It is now recognised that serratedpolyps including sessile serrated adenomas (SSAs)

1Department of LaboratoryMedicine, University ofWashington, Washington, USA2Clinical Research Division, FredHutchison Cancer ResearchCenter, Washington, USA3Department of Medicine,University of Washington,Washington, USA

Correspondence toWilliam M Grady, FredHutchinson Cancer ResearchCenter, 1100 Fairview AvenueNorth, Box 19024, D4-100,Seattle, WA 98109, USA;[email protected]

Pritchard CC, Grady WM. Gut (2010). doi:10.1136/gut.2009.206250 1 of 14

Recent advances in basic science Gut Online First, published on October 4, 2010 as 10.1136/gut.2009.206250

Copyright Article author (or their employer) 2010. Produced by BMJ Publishing Group Ltd (& BSG) under licence.


http://gut.bmj.com/


and traditional serrated adenomas (TSAs) also havethe potential for malignant transformation.10 11

These polyps are an alternative pathway to malig-nancy whereby a subset of hyperplastic polypsprogress to serrated neoplasms (SSAs or TSAs) anda fraction of these serrated neoplasms progress tocancer. Premalignant serrated polyps morefrequently arise in the proximal colon12 and areassociated with microsatellite instability and aber-rant DNA methylation at CpG islands, whereasconventional tubular adenomas arise via biallelicinactivation of the APC tumour-suppressor geneand display chromosome instability.13 Furthermore,other molecular lesions, such as BRAF V600Emutations, are characteristically found more often

in tumours arising via the serrated neoplasiapathway.13

Genomic and epigenomic instability andchromosomal alterationsGenomic and epigenomic instability distinguishesneoplastic from normal colonic epithelium and isa hallmark feature of colorectal carcinogenesis.14 15

At least four kinds of genomic or epigenetic insta-bility have been described in colorectal cancers: (1)chromosomal instability (CIN); (2) microsatelliteinstability (MSI); (3) CpG island methylatorphenotype (CIMP); and (4) global DNA hypo-methylation. Overlap between these categories andimprecise use of these terms has led to confusionand confounds interpretation of the literature.16

Thus, in this section, we will first define thedifferent types of genomic and epigenetic insta-bility in colorectal cancer and will delineate ingeneral terms how these mechanisms are clinicallyrelevant.CIN. The most common form of genomic

instability is chromosome instability, which isfound in as many as 85% of colorectal cancers.17

Chromosome instability, which can be recognisedby the presence of aneuploidy, is defined as thepresence of numerical chromosome changes ormultiple structural aberrations and is typicallyassessed by DNA flow cytometry.18 Despite thefrequent occurrence of CIN in colorectal cancer, themechanisms that give rise to this form of genomicstability and the role of aneuploidy in tumourprogression remain poorly understood. However,there is some evidence that CIN promotes cancerprogression by increasing clonal diversity.19e21

Importantly, from a clinical perspective, large meta-analyses have demonstrated that CIN is a marker ofpoor prognosis in colorectal cancers.18 22

MSI. Microsatellite unstable tumours, whichaccount for w15% of colorectal cancers, are

Table 1 Selected biomarkers that have been evaluated in colorectal cancer

Biomarker Molecular lesion Frequency in CRC Prediction Prognosis Diagnosis

KRAS Codon 12/13 activating mutations; rarelycodons 61, 117, 146

40% Yes Possible e

BRAF V600E activating mutation 10% Probable Probable Lynch syndrome

PIK3CA Helical and kinase domain mutations 20% Possible Possible e

PTEN Loss of protein by IHC 30% Possible e e

Microsatellite instability (MSI) Defined as >30% unstable loci in the NCIconsensus panel or >40% unstable loci ina panel of mononucleotide microsatelliterepeats9

15% Probable Yes Lynch Syndrome

Chromosome instability (CIN) Aneuploidy 70% Probable Yes e

18qLOH Deletion of the long arm of chromosome18

50% Probable Probable e

CpG island methylator phenotype (CIMP) Methylation of at least three loci froma selected panel of five markers

15% +/� +/� e

Vimentin (VIM) Methylation 75% e e Early Detection

TGFBR2 Inactivating mutations 30% e e e

TP53 mutations Inactivating mutations 50% e e e

APC mutations Inactivating mutations 70% e e FAP

CTNNB1 (b-catenin) Activating mutations 2% e e e

Mismatch repair genes Loss of protein by IHC; methylation;inactivating mutations

1e15% e e Lynch syndrome

CRC, colorectal cancer; FAP, familial adenomatous polyposis; IHC, immunohistochemistry.

Box 1 Summary points

< Chromosome instability (CIN) and microsatellite instability (MSI) are distinctmechanisms by which colorectal cancers arise, and that are associated withunique molecular features.

< Key pathways that drive colorectal cancer are WNT signalling, transforminggrowth factor b (TGFb) signalling and epidermal growth factor receptor(EGFR) signaling. Ras/Raf/MAPK and phosphatidylinositol 3-kinase (PI3K)pathways are both stimulated by EGFR. Currently, only downstreammediators of EGFR have a clinical role as biomarkers.

< KRAS mutations in codon 12/13 are a highly validated predictive marker forresistance to monoclonal antibody drugs that target EGFR; BRAF V600Emutation is likely to be a second predictive marker. Additional resistancemarkers including PIK3CA mutations and PTEN protein loss are beingevaluated.

< MSI+ cancers have a better prognosis and CIN+ cancers do worse; 18q LOH+ tumours also have a worse prognosis but are frequently associated withCIN. Downstream mediators of EGFR are under study for prognostication.

< The role of colorectal cancer molecular biomarkers in clinical decision makingis likely to expand as more targeted drugs become available.

2 of 14 Pritchard CC, Grady WM. Gut (2010). doi:10.1136/gut.2009.206250

Recent advances in basic science


http://gut.bmj.com/


generally regarded as being mutually exclusive ofCIN tumours because they display a normalkaryotype and exhibit unique genetic features,although there does appear to be a subset oftumours that show both CIN and MSI.16 MSIcolorectal cancer has been defined by the presenceof at least 30% unstable loci in a panel of 5e10 loci

consisting of mono- and dinucleotide tractsselected at a National Cancer Institute consensusconference.23 Currently, many clinical laboratoriesassess MSI using a panel of five mononucleotidemarkers (BAT-25, BAT-26, NR-21, NR-24 andMONO-27) that were selected for high sensitivityand specificity.9 A subset of tumours with only10e29% unstable loci has been designated asa form of microsatellite tumours designated ‘MSI-low’. Although there is evidence that MSI-lowcancers have distinct features compared with MSI(also referred to as ‘MSI-high’, or ‘MSI-H’) andmicrosatellite stable (MSS) tumours, there isconsiderable controversy regarding whether MSI-low is a unique molecular subclass of colorectalcancer.16 17 24 Patients with colorectal cancer withMSI tumours have been shown to have a betterprognosis compared with patients with CINtumours18 25, and probably respond differently toadjuvant chemotherapy compared with patientswith MSS cancer.26e28

In contrast to CIN, the mechanisms underlyingMSI are relatively well understood and involveinactivation of genes in the DNA mismatch repair(MMR) family, either by aberrant methylation orby somatic mutation.21 Furthermore, individualswith Lynch syndrome (hereditary non-polyposiscolorectal cancer, HNPCC) almost exclusivelydevelop MSI colorectal cancers because they havegermline mutations in the MMR genes, whichinclude MLH1, MSH2, MSH6 and PMS2. Incontrast, sporadic MSI colorectal cancers mostoften have loss of MMR activity as the result ofsilencing of MLH1 by aberrant methylation.21 29 Itis also now recognised that sporadic MSI tumoursare associated with the serrated neoplasia pathwayand frequently carry BRAF V600E mutations, whilecancers resulting from germline mutations in MMRgenes (Lynch syndrome) do not have mutatedBRAF.30 31 Thus, the presence of a BRAF mutationin an MSI tumour effectively excludes the possi-bility that the tumour arose as the consequence ofLynch syndrome (figure 2).CIMP. Epigenetic instability in colorectal cancer

is manifested as both hypermethylation of genepromoters that contain CpG islands (the CpGIsland methylator phenotype, CIMP), and globalDNA hypomethylation. Mechanisms that give riseto CIMP are not yet clear, although the strongassociation between BRAF V600E mutations andCIMP colorectal cancer suggests a role for activatedBRAF in the pathogenesis of the methylatorphenotype and a link between sporadic MSI andCIMP.32 33 However, in vitro studies of mutantBRAF in colorectal cancer cell lines have notdemonstrated a direct cause and effect relationshipbetween BRAF and CIMP.34 Furthermore, althoughCIMP tumours do appear to represent a distinctsubset of colorectal cancer, the clinical utility ofthis designation is hindered by lack of a universallyaccepted definition of the methylator phenotype.CIMP is usually defined as methylation of at leastthree loci from a selected panel of five gene-asso-ciated CpG islands. Because this panel is not

Figure 2 Testing strategies for Lynch syndrome (hereditary non-polyposis colorectalcancer (HNPCC)). A multistage approach to facilitate the cost-effective diagnosis ofLynch syndrome is outlined. Patients with a high clinical suspicion of Lynch syndromeare first screened by immunohistochemistry (IHC) studies of the tumour tissue to assessfor loss of mismatch repair (MMR) protein expression and by MSI testing of the tumourDNA (Tier 1 Screening Tests). Patients with tumours that show microsatellite instability(MSI) with loss of MSH2, MSH6 or PMS2 by IHC undergo germline DNA mutationanalysis of the gene corresponding to the missing protein. In contrast, patients with MSItumours that lack MLH1 are further assessed, with assessment of the tumoir for MLH1promoter methylation and mutant BRAF V600E (Tier 2 Screening Test) because mostsporadic MSI colon cancers have methylated MLH1 and Lynch syndrome MSI cancersrarely harbour BRAF mutations. When there is no evidence of MLH1 promotermethylation or BRAF mutation, mutation analysis of the MLH1 gene is performed toidentify patients with Lynch syndrome with mutations in this gene.

Figure 1 The adenomaecarcinoma progression sequence. Colorectal carcinogenesisprogresses by at least two well-recognised pathways. The chromosome instability (CIN)pathway is characterised by classic tubular adenoma histology and the early acquisitionof APC mutations that lead to deregulated WNT signalling, frequent activating mutationsof the KRAS oncogene at the early adenoma stage, loss of heterozygosity atchromosome 18q (18qLOH) in late adenomas and TP53 mutations that facilitate thetransition to frank malignancy. In contrast, tumors that harbour microsatellite instability(MSI) frequently acquire BRAF mutations and are not associated with 18qLOH or TP53mutations. Sporadic MSI cancers appear to arise commonly via the serrated neoplasiapathway, in which sessile serrated adenomas are the most frequently observedprecancerous lesions.




http://gut.bmj.com/


always the same across studies, attempts are beingmade to facilitate standardisation of CIMPmarkers for clinical use.33 35 Some authors haveproposed two classes of CIMP, CIMP-low andCIMP-high, depending on the number of methyl-ated marker loci detected.32 Another groupsuggested that CIMP colorectal cancers be dividedinto two distinct classes (called CIMP1 andCIMP2) based on the results of unsupervisedcluster analysis of a large panel of methylationmarkers.36 Finally, considerable overlap betweenCIMP and sporadic MSI tumours adds to thechallenge of incorporating CIMP status into clin-ical trials and clinical decision making.33 Retro-spective studies suggest CIMP will ultimately beshown to be a predictive marker for colorectalcancer, but the data are not adequate at this timeto recommend its clinical use.36 37 Thus, thediscovery and classification of CIMP tumours hasadvanced our understanding of the molecularpathology of colorectal cancer but has not yetimpacted clinical care.In addition to aberrant gene methylation,

a global decrease in methylation has also beenidentified in many colorectal cancers and is tightlyassociated with CIN tumours.38 39 Further researchis necessary to determine if measurement of globalDNA hypomethylation in colorectal cancer has anyrole in the clinical setting.

Role of specific genetic alterations and signalpathway deregulation as biomarkersJust as important as genomic and epigenomicinstability for the pathogenesis of colorectal canceris the accumulation of mutations in specific genesand the resulting deregulation of specific signallingpathways that control the hallmark behaviours ofcancer: cell proliferation, differentiation, apoptosis,immortalisation, angiogenesis and invasion. Thebest-studied pathways that are deregulated incolorectal cancer are the WNTeb-catenin signallingpathway, the transforming growth factor b (TGFb)signalling pathway, the EGFRemitogen-activated

protein kinase (MAPK) pathway and the phos-phatidylinositol 3-kinase (PI3K) pathway.5 16

Selected deregulated pathways in colorectal cancerand targeted treatments in clinical use or in clinicaltrials are summarised in table 2.Key tumour-suppressor genes that do not neces-

sarily mediate their effects through signal pathwayderegulation, such as TP53, and recurrent cytoge-netic aberrations such as 18q loss of heterozygosity(LOH) are also well studied in colorectal cancer andaffect the malignant transformation of colonepithelial cells through specific effects on thebehaviour of the cells (figure 1). The use of thesemolecular alterations in the management ofpatients with colorectal cancer will also bediscussed in more detail below.

WNT pathwayMutations in the adenomatous polyposis coli (APC)gene occur in up to 70% of sporadic colorectalcancers and are the cause of the familial adenoma-tous polyposis (FAP) cancer predispositionsyndrome. APC mutations can be found at theearliest stages of neoplasia and are predominantlyassociated with the classic tubular adenomapathway and CIN cancers (figure 1).6 40 41 The APCprotein negatively regulates WNT signalling viatargeting b-catenin for ubiquitin-mediated protea-somal degradation. Disruption of the APC proteinresults in increased WNT signalling through stabi-lisation of nuclear b-catenin. Activating mutationsin the b-catenin gene (CTNNB1) that protect theprotein from APC-mediated degradation are alsoobserved in colorectal neoplasia, although they arefound more frequently in adenomas (12.5%) thaninvasive cancer (1.4%), suggesting that CTNNB1-mutant tumours do not frequently progress tocarcinoma.42 Despite the critical and nearlyuniversal role of WNT pathway activation incolorectal carcinogenesis, there is currently noclinical use for APC or CTNNB1 mutations fortreatment selection, prognosis or early cancerdetection (table 1). There has been intense effort todevelop small molecule inhibitors of this pathway,but these efforts are still confined to the preclinicalarena. If these agents eventually reach the clinic,the assessment of APC mutations or activatedb-catenin (by the detection of nuclear localisationof b-catenin by immunostaining) is likely to havea role in directing the selection of patients who willrespond to these agents.

TGFb pathwayDeregulation of TGFb signalling, which is gener-ally considered a tumour-suppressor pathway inthe colon, occurs in the majority of colorectalcancers.43 Inactivating mutations have beenobserved in receptor genes (TGFBR2 and TGFBR1),postreceptor signalling pathway genes (SMAD2,SMAD4) and TGFb superfamily members(ACVR2).17 44e46 Functionally significantmutations in TGFBR2 are detected in as many as30% of all colorectal cancer and are associated withthe malignant transformation of late adenomas.

Table 2 Pathways commonly deregulated in colorectal cancer and targeted drugs inclinical use (in bold) or in clinical trials

Pathway Specific target Drugs

EGF/MAPK EGFR (mAb) Cetuximab, Panitumumab

EGFR (TKI) Erlotinib, Gefitinib

KRAS Tipifarnib, Lonafarnib

BRAF Sorafenib, PLX4032, XL281

MEK Selumetinib

PI3K PI3K BKM120, BGT226, XL147, GDC-0941

mTOR Everolimus, XL765

AKT Perifosine

WNT Resveratrol

TGFb TGFb2 AP 12009

VEGF VEGF Bevacizumab

VEGFR Vatalanib, AMG706, Pazopanib, Cediranib

HGF HGF mAb AMG102

IGF IGF-1 mAb AMG479, IMC-A12

EGF, epidermal growth factor; HGF, hepatocyte growth factor; IGF, insulin-like growth factor; mAb, monoclonalantibody; MAPK, mitogen-activated protein kinase; PI3K, phosphatidylinositol 3-kinase; TGFb, transforminggrowth factor b; TKI, tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor.




http://gut.bmj.com/


TGFBR2 mutations are most common in MSItumours, but also occur in w15% of MSS tumours(figure 1).46e48 SMAD4 is located on 18q in theregion commonly deleted in colorectal cancer, andis associated with adenoma formation and adeno-maecarcinoma progression in mouse models,supporting a role for SMAD4 as a tumour-suppressor gene.49 Furthermore, loss of SMAD4expression as detected by immunostaining hasbeen reported in >50% of colon cancers and isassociated with lymph node metastases.50 There isstill not any definite clinical role for any geneticmarkers in the TGFb signalling pathway; however,there is some evidence that SMAD4 expressionlevels may be associated with prognosis andresponse to 5-fluorouricial (5-FU), and there isongoing investigation of 18qLOH as a predictivemarker, which is discussed further in the nextsection.51 52

18qLOHLoss of the long arm of chromosome 18 (18qLOH)is the most frequent cytogenetic alteration incolorectal cancer and is observed in up to 70% oftumours.6 22 Two genes thought to have a role inthe tumourigenic effects of this loss are deleted incolorectal carcinoma (DCC) and SMAD4. Addi-tional mediators of the TGFb pathway, includingSMAD2 and SMAD7, are also in the 18qLOHregion, suggesting that 18qLOH promotestumourigenesis at least in part through deregula-tion of TGFb signalling. It appears that deletion at18q is associated with a worse prognosis; however,efforts to definitively link 18qLOH to prognosis arelimited by a lack of consistent results across studiesand heterogeneous detection methods.22 Ongoingclinical trials (eg, NCT00217737, also designatedECOG 5202) are assessing the utility of 18qLOH fortreatment selection.

Figure 3 Mediators of epidermal growth factor receptor (EGFR) signalling and anti-EGFR antibodies. EGFR formsa homodimer after ligand activation, which results in phosphorylation/activation of the intracellular kinase domain anda cascade of downstream signalling including activation of the Ras/Raf/MAPK and phosphatidylinositol 3-kinase (PI3K)pathways that are associated with cell growth, differentiation, survival and invasion. Monoclonal antibodies used totreat patients with metastatic colorectal cancer including cetuximab and panitumumab bind to the extracellular portionof EGFR and inhibit signalling in some patients. Activating mutations in KRAS occur in w40% of colorectal cancers andare thought to confer resistance to these drugs by bypassing the need for upstream EGFR signals. Activating mutationsin BRAFdthe direct downstream effector of KRASdoccur in w10% of colorectal cancers and also probably conferresistance to anti-EGFR monoclonal antibodies. Emerging evidence supports an additional role for oncogenic aberrationsin the PI3K pathway in cetuximab and panitumumab resistance.




http://gut.bmj.com/


TP53Mutations in the tumour-suppressor gene TP53occur in about half of all colorectal cancers andpromote the malignant transformation ofadenomas (figure 1).6 Like APC, TP53 is a keytumour suppressor that has been extensivelystudied in colorectal cancer but currently has nopredictive or prognostic role in the clinical setting.16

Mediators of EGF signallingEGFR/RAS/RAF/RAF/MAPKKRAS, a member of the RAS family of proto-onco-genes, is the most frequently mutated gene in all ofhuman cancer and arguably the most clinicallyimportant oncogene in colorectal cancer. The KRASprotein is a downstreameffector of EGFR that signalsthrough BRAF to activate the MAPK pathway andpromote cell growth and survival (figure 3). Muta-tions in KRAS codons 12 or 13 occur in w40% ofcolorectal cancers and lead to constitutive signallingby impairing the ability of GTPase-activatingproteins to hydrolyse KRAS-bound GTP.53 KRASmutations occur after APC mutations in the adeno-maecarcinoma progression sequence, but are stilla relatively early event in tumourigenesis (figure 1).6

Acquired KRAS mutations are maintainedthroughout carcinogenesis, as evidenced by thenearly perfect concordance of KRASmutation statusin primary andmetastatic colorectal cancer.54 55 Thisfact is critical to the utility of KRAS mutationalanalysis on archived primary tumour specimens inpatients with metastatic disease and usually elimi-nates the need for additional biopsy tissue.The BRAF gene, mutated in w10e15% of colo-

rectal cancers, encodes a protein kinase that is thedirect downstream effector of KRAS in the Ras/Raf/MAPK signalling pathway. The majority ofBRAF mutations are a single base change resultingin the substitution of glutamic acid for valine atcodon 600 (V600E; sometimes referred to as‘V599E’).5 KRAS and BRAF mutations are mutuallyexclusive, supporting the hypothesis that an acti-vating mutation in either gene is sufficient topromote tumourigenesis via increased MAPKsignalling.56 As discussed above, BRAF mutationsare much more frequent in MSI tumours (w50%)compared with MSS tumours (w5%) and are verytightly linked to CIMP cancers and the serratedneoplasia pathway.56 57 Emerging evidencesupports a role for BRAF as a genetic marker forprediction, prognosis and risk stratification.Alterations in EGFR ligands and the EGFR gene

itself are also observed in a subset of colorectalcancers. There are some data to support thatupregulation of the EGFR ligands epigregulin andamphiregulin are associated with an anti-EGFRdrug response.58 59

The PI3K pathwayMutations in PI3K pathway genes are observed inup to 40% of colorectal cancer and are nearlymutually exclusive of one another.60 The mostfrequent mutations of the PI3K pathway occur inthe p110a catalytic subunit PIK3CA, which are

reported in up to 32% of colorectal cancers and maypromote the transition from adenoma to carcinoma(figure 1).61 Mutations are also observed in PTEN,a tumour-suppressor gene that negatively regulatesPI3K signalling in as many as 30% of MSI tumoursand 9% of CIN tumours.62 The PI3K pathway ismodulated by EGFR signalling in part via KRASactivation, and there is a plausible role for bothPIK3CA and PTEN mutations as predictive markersof anti-EGFR treatment (figure 3).63 64 Currently,there is not sufficient evidence from clinical studiesto support the use of PI3K pathway mutations aspredictive or prognostic biomarkers.

RISK STRATIFICATION AND EARLY DETECTIONOne use of molecular markers in the management ofcolorectal cancer is in risk stratification for identi-fying individuals at high risk for developing colo-rectal cancer and for the early detection of colonadenomas and early-stage colorectal cancers. Withregard to risk stratification, the most robust molec-ular markers to date are germline mutations in genesthat cause the hereditary colon cancer syndromes(eg, APC mutations and FAP, BMPR1A and juvenilepolyposis, etc.) and MSI tumour status, which is anindicator of the possibility of Lynch syndrome. Theuse of MSI tumour testing in the diagnosis of Lynchsyndrome will be discussed below in the context ofMSI testing being a risk stratification markerbecause of its association with Lynch syndrome.

Lynch syndrome/HNPCC syndromeIdentifying individuals with Lynch syndrome (alsoknown as HNPCC) dramatically alters their clinicalmanagement, and can lead to effective colorectalcancer prevention programmes for these individ-uals and their family members. However, currentlythe definitive molecular diagnosis of Lynchsyndrome requires expensive germline DNAmutation analysis of multiple DNA MMR genes(MLH1, MSH2, MSH6 and PMS2). To facilitate themost cost-effective strategies for identifyingpatients at high risk for Lynch syndrome who arecandidates for genetic testing, the evaluation ofmolecular features of colorectal cancers that haveoccurred in these individuals or other familymembers can be used to predict the likelihood ofidentifying a germline mutation in one of theMMR genes. It is now common practice formolecular diagnostics laboratories to offer a step-wise series of molecular tests that are used toidentify colorectal cancers that probably arose inthe setting of Lynch syndrome. These tests arebased on the molecular pathology of colorectalcancer (figure 2).65 A common approach is initiallyto test the tumours for loss of MMR gene products(MLH1, MSH2, MSH6 and PMS2) by immuohis-tochemistry (IHC) and for MSI by PCR as the first-tier screening test (see, for example, http://www.mayomedicallaboratories.com/test-catalog/Clinical+and+Interpretive/17073), although there issupport for the use of IHC alone as a first-linetest.66 Tumours that display MSI and loss of MLH1




http://gut.bmj.com/


protein expression by IHC are then subjected toreflex testing for BRAF V600E mutation status andMLH1 promoter hypermethylation to help distin-guish sporadic MSI tumours (w50% BRAF-mutantand 99% MLH1-methylated) from Lynch syndromeMSI tumours (BRAF-WT, infrequent MLH1 meth-ylation) (figure 2, table 3).30 67 68 This strategy ismost effective in excluding individuals who areunlikely to have an MMR gene mutation fromundergoing germline mutation testing. It is notablethat in those tumours that have MSI and loss ofMSH2, MSH6 or PMS2 the likelihood of havinga germline mutation is extremely high. Also ofinterest, it is now recognised that a strategy thatrelies on clinical criteria alone for the diagnosis ofindividuals at risk for Lynch syndrome under-diagnoses this syndrome.69 In light of thesubstantial effect of a missed diagnosis on anindividual’s likelihood of developing cancer in thefuture, a strategy that employs universal testing ofall colorectal cancer is being advocated by someexperts in this area.70 It remains to be determined ifthis strategy is cost-effective and if the benefitsoutweigh the risks.

Molecular markers and early detection ofcolorectal cancerColonoscopy is the most accurate test currently forcolorectal cancer screening; however, it is expensiveand associatedwith procedure-related complicationsand poor patient compliance. In contrast, anothercommonly used colorectal cancer screening test,faecal occult blood testing (FOBT) is inexpensive andsimple to perform, but has a relatively low sensi-tivity and specificity.71 Advances in our under-standing of the molecular pathology of colorectalcancer have led to the identification of promisingearly detection molecular markers for use in non-invasive colorectal cancer screening assays.72 73

Stool-based methylated VIMENTIN (mVim) isa clinically validated marker for early colorectalcancer detection that is now commercially availablein the USA (table 1).74 The test relies on the fact thata majority of colorectal cancers (53e84%) carry anaberrantlymethylated vimentin (VIM) gene. A PCR-based assay that simultaneously measured mVimand DNA integrity reported a sensitivity of 83% anda specificity of 82%, with approximately equalsensitivity in patients with stage IeIII colorectal

cancer .75 At this time, methods are under develop-ment to enhance the performance of stool- andplasma-based methylation assays for clinicalpurposes.76 The use of molecular assays, such as thefaecal methylated VIM assay, in the clinical care ofpatients is an area that is likely to undergo rapidadvances in the near future.

GENETIC MARKERS AND PROGNOSISGenomic instability and prognosis: MSI versus CINMeta-analyses across a diverse range of patientshave firmly established that MSI colorectal cancerhave a better prognosis and that CIN tumours havean unfavourable prognosis (table 4).18 25 Thecombined HR for MSI colorectal cancers for overallsurvival (OS) was estimated to be 0.65 (95% CI0.59 to 0.71) with only one of 32 included studiesreporting an HR >1.0.25 Conversely, the overall HRassociated with CIN colorectal cancer was deter-mined to be 1.45 (95% CI 1.27 to 1.45) based on 63eligible studies and >10 000 patients.18 Despite theclear association of MSI and CIN with prognosis,these markers have not yet been adopted intoroutine clinical decision making. It is most likelythat MSI testing will be adopted into clinicalpractice before CIN testing because of the avail-ability of a reliable assay for assessing MSI status.

18qLOH and prognosisPatients with colorectal cancer with 18qLOHappear to have a worse prognosis compared withpatients with tumours that do not carry 18qLOH.A meta-analysis of 17 independent studies that waslimited by evidence of publication bias found anoverall HR of 2.00 (95% CI 1.49 to 2.69) for18qLOH across all patients, and an HR of 1.69(95% CI 1.13 to 2.54) in the adjuvant setting.22

Candidate genes in the 18q region, including DCCand SMAD4, have been studied individually forprognostic roles, with inconsistent results.77 Theindependent prognostic contribution of 18q dele-tion in colorectal cancer has been called into ques-tion due to the tight association between 18qLOHand CIN, and the inverse association of 18qLOHand MSI.16 This assertion is supported by a recentstudy that found no difference in prognosis attrib-utable to 18qLOH in a prospectively collectedcohort of 555 non-MSI tumours stage IeIV.78 Thus,it is unclear at this time if 18qLOH represents anindependent prognostic marker, or is merelya surrogate marker for CIN/MSS colorectal cancers.

Mediators of EGFR and prognosisSeveral recent studies have assessed the prognosticsignificance of KRAS, BRAF and PIK3CA mutationsin colorectal cancer.24 79e83 Mutant KRAS was notindependently associatedwith differences in relapse-free survival or OS in stage II or III colorectal cancer,but mutant BRAF was prognostic for OS in thisgroup of patients.24 79 In contrast,mutantKRAS andBRAF have been reported as markers of poor prog-nosis in advanced colorectal cancer. In the largeststudy that has addressed the prognostic role ofKRAS

Table 3 Biomarkers used in the diagnosis of Lynch syndrome (HNPCC)

Biomarker

Frequency

Sporadic Lynch syndrome

Microsatellite instability (MSI) 15% >95%

BRAF V600E mutations 50% of sporadic MSI <1%

5% of MSS

10% overall

Mismatch repair protein loss by IHC 10e15%, mostly MLH1 w90%

MLH1 promoter hypermethylation w99% of sporadic MSI <1%

<1% MSS

15% overall

HNPCC, hereditary non-polyposis colorectal cancer; IHC, immunohistochemistry; MSS, microsatellite stable.




http://gut.bmj.com/


mutations in advanced colorectal cancer to date,patients with mutantKRAS cancers had a worse OS(HR 1.40; 95% CI 1.20 to 1.65) but a similarprogression-free survival (PFS) compared withpatients with tumours bearing wild-type (WT)KRAS.81 The potential prognostic value of KRASmutations is of particular interest in advancedcolorectal cancers because the KRAS mutationalstatus of tumours is now being routinely collected inthis setting in order to assess for eligibility fortreatment with cetuximab or panitumumab. At thistime, the use ofKRASmutation status for prognosisin colorectal cancer is still premature but appears tohave significant potential to be adopted into clinicaluse in the near future.

PREDICTIVE BIOMARKERSAlthough the treatment of colorectal cancer stillprimarily relies on the surgical resection of theprimary tumour to achieve a cure, considerableprogress in the medical treatment of stage III andIV colorectal cancer has occurred over the last15 years. The adjuvant therapy of stage III colo-rectal cancer has become more effective as thestandard regimen has advanced from 5-fluorouracil(5-FU) and leucovorin to 5-FU and oxaliplatin or

irinotecan.84 Furthermore, the treatment ofpatients with stage IV colorectal cancer hasexpanded to include targeted treatments (cetux-imab, panitumumab, bevacizumab; see table 2) inaddition to 5-FU, oxaliplatin and irinotecan. Withthe identification of multiple effective agents forthe treatment of colorectal cancer has come a needfor predictive markers for selecting optimal treat-ment regimens for patients. This is particularlyapplicable to colorectal cancer because of theheterogeneity in response among colon cancers andbecause of the toxicity and cost of the medicaltreatments. The potential of genetic and epigeneticalterations to be effective predictive molecularmarkers has received considerable attention latelyand has led to the use of some of these markers inthe routine care of patients with colorectal cancer(table 5).The advent of cancer therapeutics that target

specific molecules and pathways highlights thepotential for underlying genetic and epigeneticlesions in colorectal cancer to guide personalisedtreatment decisions. A clear demonstration of thepotential of mutant genes to direct treatment isthat of mutant KRAS and treatment with cetux-imab. Only w15% of patients with metastaticcolorectal cancer respond to monoclonal antibody(mAb) therapies targeting the EGFR, whichprompted intense research into resistance mecha-nisms that could be secondary to alterations in theEGFR gene and/or mutations in downstreameffectors. These studies have produced one well-validated and exceedingly robust predictive marker(mutant KRAS) and several more promisingbiomarkers that require further validation (mutantBRAF, PIK3CA and PTEN).85 Research efforts arealso focused on identifying molecular features ofcolorectal cancer that predict response to adjuvantchemotherapy with cytoxic agents: 5-FU, irino-tecan and oxaliplatin.16 In this section we willdiscuss genetic features of colorectal cancer thathave been evaluated for a role in guiding treatmentselection. We have focused primarily on acquiredtumour mutations as predictive markers, but it is

Table 4 Prognostic biomarkers in colorectal cancer

BiomarkerMutationfrequency Prognosis Evidence Status

Microsatellite instability(MSI)

15% Favourable Strong Testing available but notyet widely used

Chromosome instability(CIN)

70% Unfavourable Strong No readily available test,not in clinical use

18qLOH/SMAD4 loss 50% Unfavourable Moderate No readily available test,not in clinical use

BRAF V600E mutations 10% Probably unfavourable Moderate Testing available butinsufficient evidenceto use for prognosis

KRAS codon 12/13mutations

40% Probably unfavourablein advanced disease

Limited Testing widely availablebut insufficient evidenceto use for prognosis

PIK3CA mutations 20% Possibly unfavourable Limited No readily available test,not in clinical use

Table 5 Colorectal cancer biomarkers as predictors for drug selection

BiomarkerMutationfrequency Drug selection Evidence Status

KRAS codon 12/13 mutations 40% Predicts resistance to anti-EGFR therapy Strong Validated, in routine clinical use

KRAS codon 61/117/146 mutations 1% Probably predicts resistance to anti-EGFRtherapy

Moderate In clinical use, not fully validated

BRAF V600E mutations 10% Probably predicts resistance to anti-EGFRtherapy, may predict response to BRAFinhibitors

Moderate In clinical use, not fully validated

PIK3CA mutations 20% May predict resistance to anti-EGFRtherapy

Limited No readily available test, not in clinicaluse

PTEN loss 30% May predict resistance to anti-EGFRtherapy

Limited No readily available test, not in clinicaluse

Microsatellite instability (MSI) 15% May predict adverse outcome with 5-FUand improved outcome with Irinotecan

Moderate Not yet in routine clinical use asa predictive biomarker

18qLOH/SMAD4 loss 50% May predict resistance to 5-FU Moderate No readily available test, not in clinicaluse

Topo1 low 50% May predict resistance to irinotecan Limited No readily available test, not in clinicaluse

EGFR, epidermal growth factor receptor; 5-FU, 5-fluorouracill LOH, loss of heterozygosity.




http://gut.bmj.com/


important to note that inherited (germline) poly-morphisms also influence the effects of chemo-therapy on cancers and the risk for drug toxicity,particularly in the case of 5-FU and irinotecan(reviewed in Walther et al16).

Predictors of response to anti-EGFR mAbtreatmentsEGFR-targeted monoclonal antibodies cetuximab(Erbitux), and the fully humanised mAb panitu-mumab (Vectibix), have proven to be effective inpatients with metastatic colorectal cancer both assingle agents and in combination with traditionalchemotherapy.86e88 However, while these treat-ments improve both PFS and OS, they are effectivein only a minority of patients with metastaticcolorectal cancer.85 These drugs are generally welltolerated, but are still associated with treatment-related morbidity, including skin rash, diarrhoeaand nausea, and are also expensive. To better targetanti-EGFR mAb treatment to patients most likelyto benefit, KRAS mutation status and additionalmolecular markers of cetuximab and panitumumabresistance have been extensively evaluated.5

KRAS is an accurate predictive biomarkerResults of four large phase III randomised trialshave established unequivocally that patients withmetastatic colorectal cancer with KRAS mutationsin codon 12 or 13 do not benefit from cetuximab orpanitumumab treatment.4 89e91 Prior to the publi-cation of these pivotal trials, the link betweenKRAS mutation status and anti-EGFR mAbresponse was already firmly supported by severalsmaller studies,92e94 but the data were not suffi-cient to warrant routine clinical testing. Therecently published randomised trials have estab-lished the use of KRAS mutational analysis asa predictive marker for anti-EGFR mAb resistancein patients with metastatic colorectal cancer inmost of the relevant clinical settings. These settingsinclude the use of cetuximab or pantumimab incombination with conventional cytotoxic chemo-therapy (eg, 5-FU, FOLFOX or FOLFIRI) as first-line treatment of metastatic disease,90 95 96 and asmonotherapy in relapsed/refractory patients.4 89 91

A second relevant question related to anti-EGFRmAb treatment is whether mutant KRAS predictsan adverse outcome in the setting of these treat-ments. The reported HRs were almost exactly 1.0in a total of 348 KRAS-mutant chemotherapy-resistant or refractory cancers treated with eitherpanitumumab89 or cetuximab4 as monotherapy,confirming lack of benefit, but also suggesting noharm from anti-EGFR mAb treatment related toPFS or OS in this population. In contrast, thereported HRs were usually >1.0 in studies ofcetuximab or panitumumab as first-line treatmentin combination with FOLFOX4 (fluorouracil,leucovorin and oxaliplatin) or FOLFIRI (fluoro-uracil, leucovorin and irinotecan) chemotherapy.85

The results of the OPUS trial (Oxaliplatin andCetuximab in First-Line Treatment of MetastaticColorectal Cancer) in particular suggest it may be

harmful to add anti-EGFR mAb treatments to 5-FU, leukovorin and oxaliplatin in patients withKRAS-mutant metastatic colorectal cancer.90

Based on the evidence from large trials, Europeanand US practice guidelines either recommend orrequire KRAS mutational analysis on colorectalcancer tumour tissue prior to the initiation ofcetuximab or panitumumab treatment.1e3 TheEuropean health authority confines use of panitu-mumab monotherapy, and cetuximab as mono- orcombination therapy, to patients with metastaticcolorectal cancer who are found to carry non-mutated WT KRAS in the primary tumours.1 TheAmerican Society for Clinical Oncology recentlypublished a provisional opinion stating that ‘Allpatients with metastatic colorectal carcinoma whoare candidates for anti-EGFR antibody therapyshould have their tumors tested for KRAS [codon 12and 13] mutations.and [KRAS-mutant] patientsshould not receive anti-EGFR antibody therapy ’.3

Similarly, the National Comprehensive CancerNetwork (NCCN) guidelines require evidence ofWTKRAS prior to cetuximab or panitumumab treat-ment in all metastatic colorectal cancer settings.2

Despite the nearly perfect negative predictivevalue of mutant KRAS, it is still only a minority(w30%) of KRAS codon 12/13 WT patients whorespond to anti-EGFR mAb treatment.85 This hasled to research into additional biomarkers thatmight predict lack of benefit in those individualswith tumours that have WT KRAS. There isevidence that rare KRAS mutations in codons 61 or146 (w2% of colorectal cancer) behave similarly tocodon 12/13 mutations,97 but incorporating thesemutations into routine clinical practice will requireanalysis of a larger group of patients. Other prom-ising markers of anti-EGFR mAb resistance areBRAF V600E mutations, PIK3CA mutations andloss of PTEN protein expression.5

BRAF: another predictor of the anti-EGFR mAbresponse?The biological rationale for BRAF V600E mutationsas an additional biomarker of anti-EGFR mAbresistance is strong: (1) BRAF is the immediatedownstream effector of KRAS in the Ras/Raf/MAPKsignalling pathway (figure 3); and (2) BRAF V600Eactivatingmutations are 100%mutually exclusive ofKRAS mutations in colorectal cancer, implying thatactivation of either protein is sufficient for colontumorigenesis. Existing limited data support BRAFV600E mutations as a negative predictor of responseto anti-EGFR mAb treatment, leading to theevolving use of BRAFmutation testing in KRAS-WTpatients prior to treatment as a means to stratifypatients further into responders and non-responders.A retrospective analysis showed that 0/11 tumourswith mutant BRAF responded to cetuximab orpanitumumab comparedwith 22/68 (32%) of BRAF-WT/KRAS-WT patients.98 Similar results wereobserved for patients treated with cetuximab plusirinotecan. None of the patients with tumours withmutant BRAF (n¼13) responded compared with 24/74 (32%) patients with tumours with BRAF-WT/




http://gut.bmj.com/


KRAS-WT.97 These finding were supported by workpresented at the 2009 American Association ofCancer Research and American Society of ClinicalOncology annual meetings,85 although not allstudies have found as robust a relationship betweenBRAF V600E mutation status and anti-EGFR anti-body response.82 99 BRAF mutations also appear tobe associated with worse prognosis independent oftreatment, which can confound the assessment ofits role as a predictive marker for response to EGFR-directed treatments.82 99 Despite the currentlylimited data, and lack of complete consensus, it islikely that BRAF mutation status has a role in anti-EGFR mAb treatment decisions and soon will beadopted into the planning for treatment withcetuximab and panitumumab.

PI3K pathway activation and anti-EGFR mAb resistanceMolecular lesions in the PI3K pathway, which incolorectal cancer are primarily mutations in PIK3CAand loss of PTEN protein expression, have beenproposed as additional anti-EGFR mAb resistancemarkers because the PI3K pathway is also stimulatedby EGFR.85 However, the relationship of oncogenicalterations in PI3K signalling and cetuximab orpanitumumab response is much less clear than thatof KRAS and BRAF mutations. In several smallstudies published to date, PIK3CA mutations orPTEN loss have been associatedwith lack of responseto cetuximab.64 100e102 Both PIK3CAmutations andPTEN loss may co-exist with KRAS or BRAF muta-tions, which weakens the biological rationale of theactivation of this pathway as an absolute predictorof anti-EGFR mAb therapeutic response. Nonethe-less, the balance of evidence points towards a prob-able predictive role formolecular events that activatethe PI3K pathway for being negative predictivemarkers for EGFR monoclonal antibody-basedtreatment. In fact, there are modest data demon-strating that when PIK3CA mutations and PTENloss of expression are combined with KRAS andBRAF mutational analysis, up to 70% of patientsunlikely to respond to cetuximab or panitumumabmay be identified.85 102 This observation has led tothe idea that colon cancer may be able to be classifiedlike breast cancers (eg, triple-negative breastcancers), and these cancers have been termed‘quadruple-negative’ for patients who do not havealterations in any of these four biomarkers.85 102

However, at this time, further studies are needed todetermine if mutant PIK3CA or PTEN loss should beincorporated into clinical practice.

EGFR mutations and amplificationThe most obvious candidate biomarker for resis-tance to antibodies which target EGFR is the EGFRgene itself. Early studies that focused on EGFRoverexpression assessed by immunohistochemistrydid not show a consistent relationship with treat-ment response, in part because of lack of stand-ardisation of the assay, which were based on eitherimmunostaining, fluorescent in situ hybridisation(FISH) or quantitative RTePCR, and interobservervariability inherent in the technique.103 EGFR gene

amplification is more promising for being a predic-tive biomarker, but has also been fraught withtechnical challenges that limit the interpretation ofexisting data, such as dilution of tumour DNAwithWT DNA in PCR-based assays, and lack of consis-tent tissue processing and scoring systems in FISHassays.5 Activating mutations in the EGFR catalyticdomain are seen frequently in lung cancer and areassociated with sensitivity to anti-EGFR tyrosinekinase inhibitors, but these mutations are quite rarein colorectal cancer.5 Thus, EGFR does not appearlikely to be a clinically useful predictive marker foranti-EGFR monoclonal antibody therapy. Further-more, although preliminary studies have shownthat the EGFR ligands amphiregulin and epiregulinare overexpressed in colorectal cancer and maypredict response to cetuximab, lack of stand-ardisation of the assays and studies that reproduc-ibly demonstrate the same effect have preventedamphiregulin and epiregulin expression levels frombeing used as clinical biomarkers for directingtreatment with EGFR mAbs.58

Predictive molecular markers for response to 5-FU,irinotecan and oxaliplatinCurrently, the tumour biomarkers that demon-strate the greatest promise for guiding adjuvantchemotherapy with conventional drugs in patientswith colorectal cancer include MSI and 18qLOH.

MSI5-FU-based regimens have been shown to be inef-fective for, or even detrimental to patients withMSI tumours.28 104 Evidence that a functioningMMR system is required for the cytotoxic effect offluorouracil provides a plausible biological rationalefor 5-FU resistance in MSI tumours.17 27 However,the finding of 5-FU resistance in MSI colorectalcancer is not uniform, and may vary with tumourstage.105 106 An ongoing phase III randomised trialof patients with completely resected stage II colo-rectal cancer (NCT00217737) will prospectivelyassess the role of MSI in predicting response toadjuvant chemotherapy in localised cancers.16

MSI tumours appear to be more responsive toirinotecan-based adjuvant chemotherapy.26

Recently published results from a large randomisedtrial of stage III colorectal cancer demonstratedimproved outcomes (both PFS and OS) in MSIpatients treated with an irinotecan-containingregimen that included 5-FU compared with 5-FU/luekovorin alone.107 In light of the prior results ofthe CALGB 98303 study showing no benefit ofadding irinotecan to 5-FU as adjuvant therapy inunselected patients with stage III colorectal cancer,the finding that MSI is a predictive biomarker foririnotecan suggests that MSI could be useful foradjusting adjuvant therapy for patients with colo-rectal cancer.108 Replication of these results inindependent studies is required to validate MSIstatus as an inclusion criterion for irinotecan-basedadjuvant chemotherapy. Currently, neither theEuropean Group on Tumour Markers nor theAmerican Society of Clinical Oncology have




http://gut.bmj.com/


recommendations on the use of MSI for guidingtreatments in patients with stage II or stage IIIcolorectal cancer.An important issue to consider with regards to

MSI is that the majority of colorectal cancers thathave MSI are sporadic colorectal cancers that haveinactivated the MLH1 gene through aberrantpromoter methylation. The majority of thesesporadic MSI tumours can be classified as CIMPcancers as well. It is not known whether theassociations seen between 5-FU and irinotecaneffects in sporadic MSI tumours also apply to MSItumours that arise in the setting of Lynchsyndrome.

Loss of 18qLoss of 18q has been associated with an adverseresponse to 5-FU-based adjuvant chemotherapy.52 109

There is some evidence that this effect is due to lossof the SMAD4 gene located in the 18q21 deletedregion, although this remains to be determinedwith more definitive studies.51 52 A number ofongoing clinical trials are assessing the predictivevalue of 18qLOH and MSI status for the treatmentof colon cancer. These include an ECOG trial ofpatients with stage II colorectal cancer being treatedwith 5-FU, oxaliplatin and bevacizumab(NCT00217737), a trial in patients being treatedwith olaparib for metastatic disease (NCT00912743),as well as a retrospective analysis assessing MSI and18qLOH in patients with colorectal cancer (stage IIor III) treated with 5-FU or 5-FU and irinotecan(CLB-9581 or CLB-89803).

Topoisomerase 1 (Topo1)In a large randomised trial that compared 5-FUalone with 5-FU + irinotecan and 5-FU + oxali-platin in advanced colorectal cancer, higherexpression of Topo1 measured by immunohisto-chemistry was significantly correlated withresponsiveness to irinotecan.110 Conversely, cancerswith low Topo1 expression (602/1269; 47%) didnot appear to benefit from the addition of irino-tecan (HR 0.98; 95% CI 0.78 to 1.22). Irinotecan isa Topo1 inhibitor; thus the level of Topo1 expres-sion has a clear biological rationale as a biomarkerfor predicting irinotecan response. Replication ofthese initial results in multiple independent studiesis required before Topo1 should be considered foruse as a predictive marker.

Polymorphisms and their role as molecular markersfor colorectal cancerWe emphasise again that germline polymorphismsthat alter pharmacokinetics and pharmacody-namics of adjuvant chemotherapy are also potentialbiomarkers for guiding treatment selection. Forexample, alterations in thymidylate synthetase anddehydropyrimidine dehydrogenase have beenextensively studied in relation to 5-FU response,and look promising. However, very few of thesepolymorphisms have been thoroughly validatedand so the majority are not ready to be used clini-cally.111 112 One exception to this generalisation is

a homozygous polymorphism that reduces theactivity of UDP-glucuronosyltransferase (UGT1A1,an enzyme that detoxifies irinotecan), which isassociated with a dose-related increased incidenceof irinotecan toxicity.113 114 This has led toa commercial UGT1A1 genotyping test that wasapproved by the Food and Drug Administration in2005 to help guide irinotecan dosing.

CONCLUSIONS AND FUTURE DIRECTIONSMore than three decades of investigations into themolecular mechanisms of colorectal cancer carci-nogenesis is finally culminating in biomarkers thatare sufficiently validated for routine clinical use.KRAS-mutational analysis to guide anti-EGFRtreatment stands as one of the first successes in theera of personalised medicine. MSI and BRAFmutations already have a clear role in triagingmolecular genetic testing in Lynch syndrome, andthese markers are poised to take on a much greaterrole in prognostication and prediction of thera-peutic responses for sporadic colorectal cancers.The use of assays for mutantKRAS, mutant BRAF

andMSI demonstrates how the molecular testing ofcolorectal cancer tissue can reduce medical costs andimprove patient outcomes by targeting therapies tothe appropriate patient population. Thus, it isanticipated that the use of molecular geneticmarkers in clinical decision making is likely toexpand as more markers are identified and validated.For example, studies are in progress assessing theefficacy of themultikinase/BRAF inhibitor sorafinib,and specific inhibitors of PI3K signalling in thetreatment of colorectal cancer.5 There is evidencethat sorafinib restores sensitivity to anti-EGFRmAbtreatment in BRAF-mutant cell lines, which hasprompted an ongoing phase II National CancerInstitute-sponsored clinical trial of sorafinib pluscetuximab in patients with metastatic colorectalcancer (NCT00343772).98 If these initial findings arevalidated, the indications for mutational analysis ofBRAF and KRAS would expand. Furthermore,patients with colorectal cancer with tumourscarrying mutant BRAF might also benefit fromnewer selective BRAF inhibitors such as PLX-4032combined with anti-EGFR mAb treatment. PIK3CAmutations or PTEN loss are likely to become clini-cally relevant for the treatment of patients withcolorectal cancer as specific PI3K pathway inhibitors(such as XL147, BGT226, GDC0941, XL765 andNVP-BEZ325) move into phase II clinical trials.115

The expanding repertoire of drugs designed toinhibit specific oncogenes and oncogenic signallingpathways again highlights that molecular mecha-nisms of colorectal cancer will increasingly playa role in the clinical care of patients with colorectalcancer. The use of molecular markers for risk strati-fication and early detection of colorectal cancer isalso showing promise, and will be part of the era ofmolecular medicine that is rapidly emerging.

Acknowledgements We thank Jonathan Tait for help withreviewing the manuscript.




http://gut.bmj.com/


Funding This work was supported by a Burroughs Welcome FundClinical Scientist in Translational Research Award, NCIRO1CA115513, and P30 CA015704 (to WMG).

Competing interests None.

Provenance and peer review Commissioned; externally peerreviewed.

REFERENCES1. European Medicines Agency. Committee for Medicinal

Products for Human Use May 2008 Plenary Meeting monthlyreport.2008. http://www.emea.europa.eu/pdfs/human/press/pr/27923508en.pdf.

2. National Comprehensive Cancer Network (NCCN) ClinicalPractice Guidelines in Oncology: Colon Cancer V.2.2010. 2010.http://www.nccn.org/professionals/physician_gls/PDF/colon.pdf.

3. Allegra CJ, Jessup JM, Somerfield MR, et al. American Societyof Clinical Oncology provisional clinical opinion: testing for KRASgene mutations in patients with metastatic colorectal carcinomato predict response to anti-epidermal growth factor receptormonoclonal antibody therapy. J Clin Oncol 2009;27:2091e6.

4. Karapetis CS, Khambata-Ford S, Jonker DJ, et al. K-rasmutations and benefit from cetuximab in advanced colorectalcancer. N Engl J Med 2008;359:1757e65.

5. Siena S, Sartore-Bianchi A, Di Nicolantonio F, et al. Biomarkerspredicting clinical outcome of epidermal growth factor receptor-targeted therapy in metastatic colorectal cancer. J Natl CancerInst 2009;101:1308e24.

6. Vogelstein B, Fearon ER, Hamilton SR, et al. Genetic alterationsduring colorectal-tumor development. N Engl J Med1988;319:525e32.

7. Dukes C. The classification of cancer of the rectum. J PatholBacteriol 1932;35:323.

8. Alexander J, Watanabe T, Wu TT, et al. Histopathologicalidentification of colon cancer with microsatellite instability.Am J Pathol 2001;158:527e35.

9. Bacher JW, Flanagan LA, Smalley RL, et al. Development ofa fluorescent multiplex assay for detection of MSI-High tumors.Dis Markers 2004;20:237e50.

10. Goldstein NS. Serrated pathway and APC (conventional)-typecolorectal polyps: molecularemorphologic correlations, geneticpathways, and implications for classification. Am J Clin Pathol2006;125:146e53.

11. Jass JR. Hyperplastic polyps and colorectal cancer: is therea link? Clin Gastroenterol Hepatol 2004;2:1e8.

12. Baker K, Zhang Y, Jin C, et al. Proximal versus distal hyperplasticpolyps of the colorectum: different lesions or a biologicalspectrum? J Clin Pathol 2004;57:1089e93.

13. Noffsinger AE. Serrated polyps and colorectal cancer: newpathway to malignancy. Annu Rev Pathol 2009;4:343e64.

14. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell2000;100:57e70.

15. Little MP, Vineis P, Li G. A stochastic carcinogenesis modelincorporating multiple types of genomic instability fitted to coloncancer data. J Theor Biol 2008;254:229e38.

16. Walther A, Johnstone E, Swanton C, et al. Genetic prognosticand predictive markers in colorectal cancer. Nat Rev Cancer2009;9:489e99.

17. Grady WM, Carethers JM. Genomic and epigenetic instability incolorectal cancer pathogenesis. Gastroenterology2008;135:1079e99.

18. Walther A, Houlston R, Tomlinson I. Association betweenchromosomal instability and prognosis in colorectal cancer:a meta-analysis. Gut 2008;57:941e50.

19. Maley CC, Galipeau PC, Finley JC, et al. Genetic clonal diversitypredicts progression to esophageal adenocarcinoma. Nat Genet2006;38:468e73.

20. Hermsen M, Postma C, Baak J, et al. Colorectal adenoma tocarcinoma progression follows multiple pathways ofchromosomal instability. Gastroenterology 2002;123:1109e19.

21. Grady WM. Genomic instability and colon cancer. CancerMetastasis Rev 2004;23:11e27.

22. Popat S, Houlston RS. A systematic review and meta-analysis ofthe relationship between chromosome 18q genotype, DCC statusand colorectal cancer prognosis. Eur J Cancer2005;41:2060e70.

23. Boland CR, Thibodeau SN, Hamilton SR, et al. A National CancerInstitute Workshop on microsatellite instability for cancerdetection and familial predisposition: development of internationalcriteria for the determination of microsatellite instability incolorectal cancer. Cancer Res 1998;58:5248e57.

24. Roth AD, Tejpar S, Delorenzi M, et al. Prognostic role of KRASand BRAF in stage II and III resected colon cancer: results of thetranslational study on the PETACC-3, EORTC 40993, SAKK 60-00trial. J Clin Oncol 2010;28:466e74.

25. Popat S, Hubner R, Houlston RS. Systematic review ofmicrosatellite instability and colorectal cancer prognosis.J Clin Oncol 2005;23:609e18.

26. Fallik D, Borrini F, Boige V, et al. Microsatellite instability isa predictive factor of the tumor response to irinotecan inpatients with advanced colorectal cancer. Cancer Res2003;63:5738e44.

27. Jo WS, Carethers JM. Chemotherapeutic implications inmicrosatellite unstable colorectal cancer. Cancer Biomark2006;2:51e60.

28. Sargent D, Marsoni S, Thibodeau SN, et al. Confirmation ofdeficient mismatch repair (dMMR) as a predictive marker for lackof benefit from 5-FU based chemotherapy in stage II and III coloncancer (CC): a pooled molecular reanalysis of randomizedchemotherapy trials. J Clin Oncol 2008;26:Abstr. 4008.

29. Kane MF, Loda M, Gaida GM, et al. Methylation of the hMLH1promoter correlates with lack of expression of hMLH1 in sporadiccolon tumors and mismatch repair-defective human tumor celllines. Cancer Res 1997;57:808e11.

30. Domingo E, Laiho P, Ollikainen M, et al. BRAF screening asa low-cost effective strategy for simplifying HNPCC genetictesting. J Med Genet 2004;41:664e8.

31. Wang L, Cunningham JM, Winters JL, et al. BRAF mutations incolon cancer are not likely attributable to defective DNAmismatch repair. Cancer Res 2003;63:5209e12.

32. Barault L, Charon-Barra C, Jooste V, et al. Hypermethylatorphenotype in sporadic colon cancer: study on a population-basedseries of 582 cases. Cancer Res 2008;68:8541e6.

33. Weisenberger DJ, Siegmund KD, Campan M, et al. CpG islandmethylator phenotype underlies sporadic microsatellite instabilityand is tightly associated with BRAF mutation in colorectal cancer.Nat Genet 2006;38:787e93.

34. Hinoue T, Weisenberger DJ, Pan F, et al. Analysis of theassociation between CIMP and BRAF in colorectal cancer by DNAmethylation profiling. PLoS One 2009;4:e8357.

35. Nosho K, Irahara N, Shima K, et al. Comprehensive biostatisticalanalysis of CpG island methylator phenotype in colorectalcancer using a large population-based sample. PLoS One 2008;3:e3698.

36. Shen L, Toyota M, Kondo Y, et al. Integrated genetic andepigenetic analysis identifies three different subclasses of coloncancer. Proc Natl Acad Sci USA 2007;104:18654e9.

37. Iacopetta B, Kawakami K, Watanabe T. Predicting clinicaloutcome of 5-fluorouracil-based chemotherapy for colon cancerpatients: is the CpG island methylator phenotype the5-fluorouracil-responsive subgroup? Int J Clin Oncol2008;13:498e503.

38. Matsuzaki K, Deng G, Tanaka H, et al. The relationship betweenglobal methylation level, loss of heterozygosity, and microsatelliteinstability in sporadic colorectal cancer. Clin Cancer Res2005;11:8564e9.

39. Rodriguez J, Frigola J, Vendrell E, et al. Chromosomal instabilitycorrelates with genome-wide DNA demethylation in humanprimary colorectal cancers. Cancer Res 2006;66:8462e9468.

40. Chung DC. The genetic basis of colorectal cancer: insights intocritical pathways of tumorigenesis. Gastroenterology2000;119:854e65.

41. Miyaki M, Iijima T, Kimura J, et al. Frequent mutation of beta-catenin and APC genes in primary colorectal tumors from patientswith hereditary nonpolyposis colorectal cancer. Cancer Res1999;59:4506e9.

42. Samowitz WS, Powers MD, Spirio LN, et al. Beta-cateninmutations are more frequent in small colorectal adenomas than inlarger adenomas and invasive carcinomas. Cancer Res1999;59:1442e4.

43. Chittenden TW, Howe EA, Culhane AC, et al. Functionalclassification analysis of somatically mutated genes in humanbreast and colorectal cancers. Genomics 2008;91:508e11.

44. Deacu E, Mori Y, Sato F, et al. Activin type II receptor restorationin ACVR2-deficient colon cancer cells induces transforminggrowth factor-beta response pathway genes. Cancer Res2004;64:7690e6.

45. Eppert K, Scherer SW, Ozcelik H, et al. MADR2 maps to 18q21and encodes a TGFbeta-regulated MAD-related protein that isfunctionally mutated in colorectal carcinoma. Cell 1996;86:543e52.

46. Grady WM, Myeroff LL, Swinler SE, et al. Mutational inactivationof transforming growth factor beta receptor type II inmicrosatellite stable colon cancers. Cancer Res 1999;59:320e4.




http://gut.bmj.com/


47. Grady WM, Rajput A, Myeroff L, et al. Mutation of the type IItransforming growth factor-beta receptor is coincident with thetransformation of human colon adenomas to malignantcarcinomas. Cancer Res 1998;58:3101e4.

48. Markowitz S, Wang J, Myeroff L, et al. Inactivation of the typeII TGF-beta receptor in colon cancer cells with microsatelliteinstability. Science 1995;268:1336e8.

49. Takaku K, Oshima M, Miyoshi H, et al. Intestinal tumorigenesisin compound mutant mice of both Dpc4 (Smad4) and Apc genes.Cell 1998;92:645e56.

50. Tanaka T, Watanabe T, Kazama Y, et al. Loss of Smad4protein expression and 18qLOH as molecular markers indicatinglymph node metastasis in colorectal cancerda studymatched for tumor depth and pathology. J Surg Oncol2008;97:69e73.

51. Alhopuro P, Alazzouzi H, Sammalkorpi H, et al. SMAD4 levelsand response to 5-fluorouracil in colorectal cancer. Clin CancerRes 2005;11:6311e16.

52. Boulay JL, Mild G, Lowy A, et al. SMAD4 is a predictive markerfor 5-fluorouracil-based chemotherapy in patients with colorectalcancer. Br J Cancer 2002;87:630e4.

53. Downward J. Targeting RAS signalling pathways in cancertherapy. Nat Rev Cancer 2003;3:11e22.

54. Artale S, Sartore-Bianchi A, Veronese SM, et al. Mutations ofKRAS and BRAF in primary and matched metastatic sites ofcolorectal cancer. J Clin Oncol 2008;26:4217e19.

55. Zauber P, Sabbath-Solitare M, Marotta SP, et al. Molecularchanges in the Ki-ras and APC genes in primary colorectalcarcinoma and synchronous metastases compared with thefindings in accompanying adenomas. Mol Pathol2003;56:137e40.

56. Rajagopalan H, Bardelli A, Lengauer C, et al. Tumorigenesis:RAF/RAS oncogenes and mismatch-repair status. Nature2002;418:934.

57. Lubomierski N, Plotz G, Wormek M, et al. BRAF mutations incolorectal carcinoma suggest two entities of microsatellite-unstable tumors. Cancer 2005;104:952e61.

58. Jacobs B, De Roock W, Piessevaux H, et al. Amphiregulin andepiregulin mRNA expression in primary tumors predicts outcomein metastatic colorectal cancer treated with cetuximab. J ClinOncol 2009;27:5068e74.

59. Lievre A, Blons H, Laurent-Puig P. Oncogenic mutations aspredictive factors in colorectal cancer. Oncogene2010;29:3033e43.

60. Parsons DW, Wang TL, Samuels Y, et al. Colorectal cancer:mutations in a signalling pathway. Nature 2005;436:792.

61. Samuels Y, Wang Z, Bardelli A, et al. High frequency ofmutations of the PIK3CA gene in human cancers. Science2004;304:554.

62. Danielsen SA, Lind GE, Bjornslett M, et al. Novel mutations ofthe suppressor gene PTEN in colorectal carcinomas stratified bymicrosatellite instability- and TP53 mutation-status. Hum Mutat2008;29:E252e62.

63. Razis E, Briasoulis E, Vrettou E, et al. Potential value of PTEN inpredicting cetuximab response in colorectal cancer: anexploratory study. BMC Cancer 2008;8:234.

64. Sartore-Bianchi A, Martini M, Molinari F, et al. PIK3CAmutations in colorectal cancer are associated with clinicalresistance to EGFR-targeted monoclonal antibodies. Cancer Res2009;69:1851e7.

65. Baudhuin LM, Burgart LJ, Leontovich O, et al. Use ofmicrosatellite instability and immunohistochemistry testing forthe identification of individuals at risk for Lynch syndrome.Fam Cancer 2005;4:255e65.

66. Shia J. Immunohistochemistry versus microsatellite instabilitytesting for screening colorectal cancer patients at risk forhereditary nonpolyposis colorectal cancer syndrome. Part I.The utility of immunohistochemistry. J Mol Diagn2008;10:293e300.

67. Bettstetter M, Dechant S, Ruemmele P, et al. Distinction ofhereditary nonpolyposis colorectal cancer and sporadicmicrosatellite-unstable colorectal cancer through quantification ofMLH1 methylation by real-time PCR. Clin Cancer Res2007;13:3221e8.

68. Zhang L. Immunohistochemistry versus microsatellite instabilitytesting for screening colorectal cancer patients at risk forhereditary nonpolyposis colorectal cancer syndrome. Part II. Theutility of microsatellite instability testing. J Mol Diagn2008;10:301e7.

69. Hampel H, Frankel WL, Martin E, et al. Screening for the Lynchsyndrome (hereditary nonpolyposis colorectal cancer). N Engl JMed 2005;352:1851e60.

70. South CD, Yearsley M, Martin E, et al. Immunohistochemistrystaining for the mismatch repair proteins in the clinicalcare of patients with colorectal cancer. Genet Med2009;11:812e17.

71. Smith RA, Cokkinides V, Brooks D, et al. Cancer screening in theUnited States: a review of current American Cancer Societyguidelines and issues in cancer screening. CA Cancer J Clin2010;60:99e119.

72. Ahlquist DA. Molecular detection of colorectal neoplasia.Gastroenterology 2010;138:2127e39.

73. Osborn NK, Ahlquist DA. Stool screening for colorectalcancer: molecular approaches. Gastroenterology2005;128:192e206.

74. Itzkowitz SH, Jandorf L, Brand R, et al. Improved fecal DNA testfor colorectal cancer screening. Clin Gastroenterol Hepatol2007;5:111e17.

75. Itzkowitz S, Brand R, Jandorf L, et al. A simplified, noninvasivestool DNA test for colorectal cancer detection.Am J Gastroenterol 2008;103:2862e70.

76. Li M, Chen WD, Papadopoulos N, et al. Sensitive digitalquantification of DNA methylation in clinical samples.Nat Biotechnol 2009;27:858e63.

77. Carethers JM, Hawn MT, Greenson JK, et al. Prognosticsignificance of allelic loss at chromosome 18q21 for stage IIcolorectal cancer. Gastroenterology 1998;114:1188e95.

78. Ogino S, Nosho K, Irahara N, et al. Prognostic significance andmolecular associations of 18q loss of heterozygosity: a cohortstudy of microsatellite stable colorectal cancers. J Clin Oncol2009;27:4591e8.

79. Fuchs C, Ogino S, Meyerhardt JA. KRAS mutation, cancerrecurrence, and patient survival in stage III colon cancer: findingsfrom CALGB 89803. J Clin Oncol 2009;27:A-4037.

80. Ogino S, Nosho K, Kirkner GJ, et al. PIK3CA mutation isassociated with poor prognosis among patients with curativelyresected colon cancer. J Clin Oncol 2009;27:1477e84.

81. Richman SD, Seymour MT, Chambers P, et al. KRAS and BRAFmutations in advanced colorectal cancer are associated with poorprognosis but do not preclude benefit from oxaliplatin oririnotecan: results from the MRC FOCUS trial. J Clin Oncol2009;27:5931e7.

82. Tol J, Nagtegaal ID, Punt CJ. BRAF mutation in metastaticcolorectal cancer. N Engl J Med 2009;361:98e9.

83. Zlobec I, Kovac M, Erzberger P, et al. Combined analysis ofspecific KRAS mutation, BRAF and microsatellite instabilityidentifies prognostic subgroups of sporadic and hereditarycolorectal cancer. Int J Cancer 2010.

84. Meyerhardt JA, Mayer RJ. Systemic therapy for colorectalcancer. N Engl J Med 2005;352:476e87.

85. Bardelli A, Siena S. Molecular mechanisms of resistance tocetuximab and panitumumab in colorectal cancer. J Clin Oncol2010;28:1254e61.

86. Ciardiello F, Tortora G. EGFR antagonists in cancer treatment.N Engl J Med 2008;358:1160e74.

87. Cunningham D, Humblet Y, Siena S, et al. Cetuximabmonotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med2004;351:337e45.

88. Saltz LB, Meropol NJ, Loehrer PJ Sr, et al. Phase II trial ofcetuximab in patients with refractory colorectal cancer thatexpresses the epidermal growth factor receptor. J Clin Oncol2004;22:1201e8.

89. Amado RG, Wolf M, Peeters M, et al. Wild-type KRAS isrequired for panitumumab efficacy in patients with metastaticcolorectal cancer. J Clin Oncol 2008;26:1626e34.

90. Bokemeyer C, Bondarenko I, Makhson A, et al. Fluorouracil,leucovorin, and oxaliplatin with and without cetuximab in thefirst-line treatment of metastatic colorectal cancer. J Clin Oncol2009;27:663e71.

91. Van Cutsem E, Peeters M, Siena S, et al. Open-label phase IIItrial of panitumumab plus best supportive care compared withbest supportive care alone in patients with chemotherapy-refractory metastatic colorectal cancer. J Clin Oncol2007;25:1658e64.

92. Benvenuti S, Sartore-Bianchi A, Di Nicolantonio F, et al.Oncogenic activation of the RAS/RAF signalling pathway impairsthe response of metastatic colorectal cancers to anti-epidermalgrowth factor receptor antibody therapies. Cancer Res2007;67:2643e8.

93. Di Fiore F, Blanchard F, Charbonnier F, et al. Clinical relevance ofKRAS mutation detection in metastatic colorectal cancertreated by cetuximab plus chemotherapy. Br J Cancer2007;96:1166e9.




http://gut.bmj.com/


94. Lievre A, Bachet JB, Le Corre D, et al. KRAS mutation status ispredictive of response to cetuximab therapy in colorectal cancer.Cancer Res 2006;66:3992e5.

95. Douillard J, Siena S, Cassidy J. Randomized phase 3 study ofpanitumumab with FOLFOX4 compared to FOLFOX4 alone as 1st-line treatment (tx) for metastatic colorectal cancer (mCRC): ThePRIME trial. Eur J Cancer 2009;7:S6:(suppl; abstr 10LBA).

96. Van Cutsem E, Kohne CH, Hitre E, et al. Cetuximab andchemotherapy as initial treatment for metastatic colorectalcancer. N Engl J Med 2009;360:1408e17.

97. Loupakis F, Ruzzo A, Cremolini C, et al. KRAS codon 61, 146and BRAF mutations predict resistance to cetuximab plusirinotecan in KRAS codon 12 and 13 wild-type metastaticcolorectal cancer. Br J Cancer 2009;101:715e21.

98. Di Nicolantonio F, Martini M, Molinari F, et al. Wild-type BRAFis required for response to panitumumab or cetuximab inmetastatic colorectal cancer. J Clin Oncol 2008;26:5705e12.

99. Laurent-Puig P, Cayre A, Manceau G, et al. Analysis of PTEN,BRAF, and EGFR status in determining benefit from cetuximabtherapy in wild-type KRAS metastatic colon cancer. J Clin Oncol2009;27:5924e30.

100. Jhawer M, Goel S, Wilson AJ, et al. PIK3CA mutation/PTENexpression status predicts response of colon cancer cells to theepidermal growth factor receptor inhibitor cetuximab. Cancer Res2008;68:1953e61.

101. Prenen H, De Schutter J, Jacobs B, et al. PIK3CA mutations arenot a major determinant of resistance to the epidermal growthfactor receptor inhibitor cetuximab in metastatic colorectalcancer. Clin Cancer Res 2009;15:3184e8.

102. Sartore-Bianchi A, Di Nicolantonio F, Nichelatti M, et al. Multi-determinants analysis of molecular alterations for predictingclinical benefit to EGFR-targeted monoclonal antibodies incolorectal cancer. PLoS One 2009;4:e7287.

103. Shia J, Klimstra DS, Li AR, et al. Epidermal growth factorreceptor expression and gene amplification in colorectalcarcinoma: an immunohistochemical and chromogenic in situhybridization study. Mod Pathol 2005;18:1350e6.

104. Ribic CM, Sargent DJ, Moore MJ, et al. Tumor microsatellite-instability status as a predictor of benefit from fluorouracil-basedadjuvant chemotherapy for colon cancer. N Engl J Med2003;349:247e57.

105. Kim GP, Colangelo LH, Wieand HS, et al. Prognostic andpredictive roles of high-degree microsatellite instability in coloncancer: a National Cancer InstituteeNational Surgical AdjuvantBreast and Bowel Project Collaborative Study. J Clin Oncol2007;25:767e72.

106. Liang JT, Huang KC, Lai HS, et al. High-frequency microsatelliteinstability predicts better chemosensitivity to high-dose5-fluorouracil plus leucovorin chemotherapy for stage IV sporadiccolorectal cancer after palliative bowel resection. Int J Cancer2002;101:519e25.

107. Bertagnolli MM, Niedzwiecki D, Compton CC, et al.Microsatellite instability predicts improved response to adjuvanttherapy with irinotecan, fluorouracil, and leucovorin in stage IIIcolon cancer: Cancer and Leukemia Group B Protocol 89803.J Clin Oncol 2009;27:1814e21.

108. Saltz LB, Niedzwiecki D, Hollis D, et al. Irinotecan fluorouracilplus leucovorin is not superior to fluorouracil plus leucovorin aloneas adjuvant treatment for stage III colon cancer: results of CALGB89803. J Clin Oncol 2007;25:3456e61.

109. Watanabe T, Wu TT, Catalano PJ, et al. Molecular predictors ofsurvival after adjuvant chemotherapy for colon cancer. N Engl JMed 2001;344:1196e206.

110. Braun MS, Richman SD, Quirke P, et al. Predictive biomarkers ofchemotherapy efficacy in colorectal cancer: results from the UKMRC FOCUS trial. J Clin Oncol 2008;26:2690e8.

111. Ezzeldin HH, Diasio RB. Predicting fluorouracil toxicity: can wefinally do it? J Clin Oncol 2008;26:2080e2.

112. Schwab M, Zanger UM, Marx C, et al. Role of genetic andnongenetic factors for fluorouracil treatment-related severetoxicity: a prospective clinical trial by the German 5-FU ToxicityStudy Group. J Clin Oncol 2008;26:2131e8.

113. Hoskins JM, Goldberg RM, Qu P, et al. UGT1A1*28 genotypeand irinotecan-induced neutropenia: dose matters. J Natl CancerInst 2007;99:1290e5.

114. Palomaki GE, Bradley LA, Douglas MP, et al. CanUGT1A1 genotyping reduce morbidity and mortality inpatients with metastatic colorectal cancer treated withirinotecan? An evidence-based review. Genet Med2009;11:21e34.

115. Yuan TL, Cantley LC. PI3K pathway alterations in cancer:variations on a theme. Oncogene 2008;27:5497e510.




http://gut.bmj.com/


colorectal cancer molecular biology moves into clinical

Documents