speciﬁc mutations in kras codons 12 and 13, and patient ...within the group of kras wild-type...

Imaging, Diagnosis, Prognosis

Specific Mutations in KRAS Codons 12 and 13, and PatientPrognosis in 1075 BRAF Wild-Type Colorectal Cancers

Yu Imamura1, Teppei Morikawa1, Xiaoyun Liao1, Paul Lochhead1, Aya Kuchiba1, Mai Yamauchi1,Zhi Rong Qian1, Reiko Nishihara1, Jeffrey A. Meyerhardt1, Kevin M. Haigis4,Charles S. Fuchs1,2, and Shuji Ogino1,3

AbstractPurpose: To assess prognostic roles of various KRAS oncogene mutations in colorectal cancer, BRAF

mutation status must be controlled for because BRAF mutation is associated with poor prognosis, and

almost all BRAFmutants are present among KRASwild-type tumors. Taking into account experimental data

supporting a greater oncogenic effect of codon 12 mutations compared with codon 13 mutations, we

hypothesized that KRAS codon 12–mutated colorectal cancers might behave more aggressively than KRAS

wild-type tumors and codon 13 mutants.

Experimental design: Using molecular pathological epidemiology database of 1,261 rectal and colon

cancers, we examined clinical outcome and tumor biomarkers ofKRAS codon 12 and 13mutations in 1,075

BRAF wild-type cancers (i.e., controlling for BRAF status). Cox proportional hazards model was used to

compute mortality HR, adjusting for potential confounders, including stage, PIK3CA mutations, micro-

satellite instability, CpG island methylator phenotype, and LINE-1 methylation.

Results: Compared with patients with KRAS wild-type/BRAF wild-type cancers (N ¼ 635), those withKRAS codon 12 mutations (N ¼ 332) experienced significantly higher colorectal cancer–specific mortality[log-rank P¼ 0.0001; multivariate HR, 1.30; 95% confidence interval (CI), 1.02–1.67; P¼ 0.037], whereasKRAS codon13–mutated cases (N¼108)were not significantly associatedwithprognosis. Among the sevenmost common KRAS mutations, c.35G>T (p.G12V; N ¼ 93) was associated with significantly highercolorectal cancer–specific mortality (log-rank P ¼ 0.0007; multivariate HR, 2.00; 95% CI, 1.38–2.90, P ¼0.0003) compared with KRAS wild-type/BRAF wild-type cases.

Conclusions: KRAS codon 12 mutations (in particular, c.35G>T), but not codon 13 mutations, areassociated with inferior survival in BRAF wild-type colorectal cancer. Our data highlight the importance of

accurate molecular characterization in colorectal cancer. Clin Cancer Res; 18(17); 4753–63. �2012 AACR.

IntroductionColorectal cancer develops through a multistep carcino-

genic process with an accumulation of epigenetic andgenetic changes, including KRAS mutation. Approximately

40%of colorectal cancers harborKRASmutations, and 90%of those mutations occur in codons 12 and 13 (1–3). Incontrast to the widely accepted predictive role of KRASmutation in identifying resistance to anti-EGFR therapy(3–8), the prognostic role of KRAS mutation in colorectalcancer remains uncertain (9–14). Recently, the differentialbiologic effect of various KRAS mutations in colorectalcancer was brought to light by data from De Roock andcolleagues (15) showing that the c.38G>A (p.G13D) muta-tion was associated with benefit from cetuximab, whereasKRAS codon 12 mutations were associated with resistanceto cetuximab among chemotherapy-refractory colorectalcancer patients. A search of the literature to-date revealsthat several studies (16–21) have compared the prognosticroles of KRAS codon 12 mutations with those of codon 13.Nonetheless, there is a lack of agreement as to the prog-nostic difference between KRAS codon 12 and codon 13mutations in colorectal cancer (Table 1).

Of note, little attention has been given to the confound-ing effect of BRAF mutation on the relationship betweenKRAS mutation and clinical outcome in colorectal cancer.

Authors' Affiliations: 1Department of Medical Oncology, Dana-FarberCancer Institute and Harvard Medical School; 2Channing Laboratory,Department of Medicine, 3Department of Pathology, Brigham andWomen's Hospital, and Harvard Medical School, Boston; and 4MolecularPathology Unit and Center for Cancer Research, Massachusetts GeneralHospital, Charlestown, Massachusetts

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Y. Imamura, T. Morikawa, X. Liao, P. Lochhead, C.S Fuchs, and S. Oginocontributed equally to this work.

Corresponding Author: Shuji Ogino, Center for Molecular OncologicPathology, Dana-Farber Cancer Institute, Brigham andWomen's Hospital,Harvard Medical School, 450 Brookline Ave., Room JF-215C, Boston, MA02215. Phone: 617-632-1972; Fax: 617-582-8558; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-11-3210

�2012 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 4753

on June 7, 2021. © 2012 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 2, 2012; DOI: 10.1158/1078-0432.CCR-11-3210

http://clincancerres.aacrjournals.org/

Almost all BRAF-mutated colorectal cancers are presentwithin the group of KRAS wild-type cancers. Comparedwith BRAF wild-type cases, BRAF mutation has been asso-ciated with poorer prognosis in several studies (10,11, 19, 21–23); hence, it is impossible to clarify the exactprognostic roles of KRAS mutations in colorectal cancerwithout controlling for BRAFmutation. Importantly, noneof the previous large studies (with a sample size ofN� 300;refs. 16, 17, 19, 20) controlled for the potential confound-ing effect ofBRAFmutation,whereas only one smaller study(N¼ 229) assessed BRAF status (ref. 21; Table 1). One wayof controlling for BRAF mutation is to examine the prog-nostic significance of KRAS mutation in BRAF wild-typecolorectal cancers. Considering experimental data (24, 25)supporting a greater oncogenic effect of KRAS codon 12mutations than codon 13 mutations, we hypothesized thatKRAS codon 12–mutated colorectal cancer might behavemore aggressively than codon 13 mutants and KRAS wild-type tumors.

We therefore tested this hypothesis by conducting a studyon the prognostic roles ofKRAS codon 12 and 13mutationsusing 1,261 colorectal cancers within 2 U.S. nationwideprospective cohort studies, in which there were 1,075 BRAFwild-type cancers. Because our molecular pathologic epi-demiology (26–28) database included tumor molecularvariables, including microsatellite instability (MSI), CpGisland methylator phenotype (CIMP), BRAF and PIK3CAmutations, and LINE-1 methylation, we could evaluate theprognostic role of KRAS codon 12 and 13 mutations inde-pendent of other potential molecular confounders. Ourfindings raise a possible need for tumor subtyping basedon specific KRAS and BRAF oncogene mutations in colo-rectal cancer.

Materials and MethodsStudy population

We used the database of 2 U.S. nationwide prospectivecohort studies, the Nurses’ Health Study (N ¼ 121,701women followed since 1976) and the Health ProfessionalsFollow-up Study (N ¼ 51,529 men followed since 1986;ref. 29). Every 2 years, cohort participants have been sentfollow-up questionnaires to identify newly diagnosed can-cers in themselves and their first-degree relatives. We col-lected paraffin-embedded tissue blocks from hospitalswhere patients underwent colorectal cancer resections(29). We collected diagnostic biopsy specimens for rectalcancer patients who received preoperative treatment, toavoid artifacts or bias introduced by treatment. Hematox-ylin and eosin (HE)-stained tissue sections from all colo-rectal cancer cases were reviewed by a pathologist (S.O.),unaware of other data. A subset of cases (N ¼ 172) werereviewed by another pathologist (T.M.), and the concor-dance between the 2 observers was 0.96 (k ¼ 0.72; P <0.0001), indicating substantial agreement. The tumor dif-ferentiation was categorized as well moderate versus poor(>50% vs. �50% gland formation). Initially, 1,261 colo-rectal cancer cases diagnosed up to 2006 were includedbased on the availability of tumor tissue, sequencing datafor both KRAS and BRAF, and survival data (Table 2).Treatment data were not available in this study. In thisstudy, BRAF-mutated cancers (N ¼ 181) were excluded toassess the prognostic role of various KRAS mutations in apool of BRAFwild-type cases. Tumors harboring mutationsin both codons 12 and 13 (N ¼ 5) of KRAS were excluded,resulting in a final total of 1,075BRAFwild-type cases as oursurvival analysis study base (Fig. 1, Supplementary TableS1). Patients were observed until death or January 1, 2011,whichever came first. Death of a participant was confirmedby searching the National Death Index. Informed consentwas obtained from all study subjects. This study wasapproved by the Human Subjects Committees at HarvardSchool of Public Health and Brigham and Women’sHospital.

Sequencing of KRAS, BRAF and PIK3CA, and MSIanalysis

Resected tissuewas fixed, processed, and stored accordingto standard protocols by hospitals across the United States,where study participants had undergone surgery. Retrievedparaffin-embedded tissue blocks were stored at room tem-perature before use. Tissue for analysis was macrodissectedfrom 10-micron sections on glass slides while guided by ahematoxylin and eosin–stained section with tumor areasmarked by a pathologist (S.O.). Accordingly, we extractedDNA from tumor tissue enriched with neoplastic cells,without adjacent normal tissue. DNA was stored at �20degrees centigrade before use. We carried out PCR andpyrosequencing targeted for KRAS (codons 12 and 13;ref. 30), BRAF (codon 600; ref. 31), and PIK3CA (exons9 and 20), as previously described (32). Pyrosequencingtechnology has been shown to reliably detect KRAS

Translational RelevanceTo assess prognostic roles of KRASmutations in colo-

rectal cancer, BRAF mutation status must be controlledbecause BRAF-mutated cancers are associated withpoorer prognosis than BRAFwild-type cases, and almostall BRAF mutants are present among KRAS wild-typetumors.However, noprevious large study (with a samplesize of N � 300) has controlled for the effect of BRAFmutation. We examined the prognostic roles of variousKRAS codon 12 and 13 mutations in 1,075 BRAF wild-type colorectal cancer cases with available clinical data,adequate follow-up, andother importantmolecular datarelevant to colorectal cancer. This study, including over1,000 BRAF wild-type colorectal cancers, shows thatKRAS codon 12 mutations (in particular, c.35G>T), butnot codon 13 mutations, are associated with inferiorsurvival, independent of clinical, pathologic, andmolec-ular features of colorectal cancer. Our data suggest thepotential need to evaluate both BRAF mutation statusand specificKRASmutation status as prognostic biomar-kers for colorectal cancer.

Imamura et al.

Clin Cancer Res; 18(17) September 1, 2012 Clinical Cancer Research4754




mutation with 100% analytic sensitivity and specificity,even when the proportion of mutant alleles is as low as10% (30). We dissected tumor-only areas, maintainingneoplastic cellularity of at least 30%. Assuming no labora-tory error, both positive and negative predictive values areestimated to be 100%.MSI analysis was carried out using 10microsatellite markers (D2S123, D5S346, D17S250,BAT25, BAT26, BAT40, D18S55, D18S56, D18S67, andD18S487; ref. 2). MSI high was defined as instability in30% or more of the markers, and MSI-low/microsatellitestability (MSS) as instability in 0% to 29% of the markers.

Methylation analyses for CpG islands and LINE-1Using validated bisulfite DNA treatment and real-time

PCR (MethyLight), we quantified DNA methylation in 8CIMP-specific promoters [CACNA1G, CDKN2A (p16),CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1;refs. 2, 33]. CIMP high was defined as the presence of 6 of 8ormoremethylated promoters, CIMP low as 1 of 8 to 5 of 8methylated promoters, and CIMP 0 as the absence (0 of 8)of methylated promoters according to the previously estab-lished criteria (2). To accurately quantify differences inrelatively high LINE-1 methylation levels, we used pyrose-quencing as previously described (34, 35).

Statistical analysisAll statistical analyses were carried out by using SAS

(Version 9.2, SAS Institute). All P values were 2-sided. Forour main hypothesis on the prognostic significance ofKRAS codon 12 mutation among BRAF wild-type cases, aP value for significance was set at P ¼ 0.05. When wecarried out multiple hypothesis testing on specific KRASmutations (the 7 most common mutations), the P valuefor significance was adjusted by Bonferroni correction toP ¼ 0.007 (¼ 0.05/7). When we carried out multiplehypothesis testing on associations or interactionsbetween KRAS mutations (codon 12 or 13) and othercovariates, a P value for significance was adjusted byBonferroni correction to P ¼ 0.0021 (¼ 0.05/24). Forcategorical data, the c2 test was done. To compare meanage and mean LINE-1 methylation level, the t test assum-ing equal variances was carried out.The Kaplan–Meier method was used to assess survival

timedistribution, and log-rank testwas used. For analyses ofcolorectal cancer–specific mortality, deaths as a result ofother causes were censored. To control for confounding, weused multivariate Cox proportional hazards regressionmodels. A multivariate model initially included sex, age atdiagnosis (continuous), year of diagnosis (continuous),family history of colorectal cancer in any first-degree relative(present vs. absent), tumor location (colon vs. rectum),tumor differentiation (well to moderate vs. poor), MSI(high vs. low/MSS), CIMP (high vs. low/0), PIK3CA, andLINE-1methylation (continuous). To avoid overfitting andresidual confounding, disease stage (I, IIA, IIB-C, IIIA, IIIB,IIIC, IV, or unknown) was used as a stratifying variableusing the "strata" option in the SAS "proc phreg" command.A backward stepwise elimination was carried out with P ¼

0.20 as a threshold to avoid overfitting. For cases withmissing information in any of the categorical covariates[tumor location (0.5%), tumor differentiation (0.7%), MSI(2.0%), CIMP (7.2%), and PIK3CA (8.5%)], we includedthose cases in the majority category of a given covariate toavoid overfitting. We confirmed that excluding cases withmissing information in any of the covariates did not sub-stantially alter results (data not shown). Theproportionalityof hazards assumption was satisfied by evaluating time-dependent variables, which were the cross-products of theKRAS indicator variables (codon 12 mutant and codon 13mutant; vs. KRAS wild-type/BRAF wild-type) and survivaltime (all P values >0.14 for colorectal cancer–specific mor-tality and overall mortality). We also tested for potentialinteraction between KRAS mutation status and each of theother covariates (including sex, age at diagnosis, familyhistory of colorectal cancer, tumor location, disease stage,tumor differentiation, MSI, CIMP, PIK3CA, and LINE-1methylation). An interaction was assessed by the Wald teston the cross-product of each of theKRAS indicator variablesand another variable of interest (without data-missingcases) in a multivariate Cox model.

ResultsKRAS mutation status in colorectal cancer

Among 1,261 patients with incident colorectal cancer inthe 2 U.S. nationwide prospective cohort studies, wedetected KRAS codon 12 and/or 13 mutations in 451(36%) patients; 335 in codon 12 only, 110 in codon 13only, and 6 in both codons 12 and 13. For KRAS codon 12mutations, we identified 161 cases with c.35G>A (p.G12D,codon 12 GGT>GAT), 95 cases with c.35G>T (p.G12V,codon 12 GGT>GTT), 44 cases with c.34G>T (p.G12C,codon 12 GGT>TGT), 20 cases with c.35G>C (p.G12A,codon 12 GGT>GCT), 12 cases with c.34G>A (p.G12S,codon 12 GGT>AGT), 8 cases with c.34G>C (p.G12R,codon 12 GGT>CGT), and one case with c.35_36delinsCA(p.G12A, codon 12 GGT>GCA). In codon 13, we identified110 cases with c.38G>A (p.G13D, codon 13 GGC>GAC), 3cases with c.37G>T (p.G13C, codon 13GGC>TGC), 2 caseswith c.38G>T (p.G13V, codon 13GGC>GTC), and one casewith c.37G>C (p.G13R, codon 13 GGC>CGC).

Table 2 summarizes the baseline characteristics of studysubjects (N ¼ 1,261) according to KRAS mutation status.There was no significant difference in any of the featuresexamined between KRAS codon 12 and codon 13 mutants.Supplementary Table S1 summarizes the baseline charac-teristics of BRAF wild-type cases (N¼ 1075) that were usedfor subsequent survival analyses.

KRAS mutation status and patient survival in BRAFwild-type cases

When assessing the prognostic effect of KRASmutation,it is necessary to consider a confounding effect of BRAFmutation because BRAF mutation is associated withpoorer prognosis (10, 11, 19, 21–23) and inversely asso-ciated with KRAS mutation (Table 2). To assess

KRAS Codon 12 and Codon 13 Mutations in Colorectal Cancer

www.aacrjournals.org Clin Cancer Res; 18(17) September 1, 2012 4755




Tab

le1.

Studieson

progn

ostic

sign

ifica

nceof

KRASco

don

12an

d13

mutations

inco

lorectal

canc

er

No.o

fKRASmutan

tsMultiva

riateHR

(95%

CI;vs

.KRAS-

Ref.

Autho

rs(y)

No.o

fho

spitals

Sam

ple

size

Tum

or

loca

tion

Disea

sestag

eco

don12

(No.o

fev

ents)

codon13

(No.o

fev

ents)

BRAF

data

wild

-typ

eas

areferent,

unless

otherwisesp

ecified

)Notes

16Sam

owitan

dco

lleag

ues(200

0)Man

y1,41

3Colon

I–IV

353(–)

100(–)

No

Can

cer-sp

ecificsu

rvival

Adjusted

byag

ean

dstag

e.Cod

on12

,1.0

(0.8–1.2)

Cod

on13

,1.4

(0.95–

2.0)

c.35

G>A

,1.1

(0.8–1.5)

c.35

G>T

,0.8

(0.5–1.2)

c.38

G>A

,1.4

(0.95–

2.0)

17And

reye

van

dco

lleag

ues

(200

1;Meta-an

alys

is)

A

,0.94(0.79–

1.11

)c.35

G>C

,1.35(0.98–

1.87

)c.35

G>T

,1.29(1.08–

1.55

)c.38

G>A

,0.93(0.78–

1.12

)18

Baz

anan

dco

lleag

ues(200

2)1

160

Colon

and

rectum

I–IV

40(21)

34(28)

No

Can

cer-sp

ecificsu

rvival

Cov

ariateswereloca

tion,

stag

e,su

rgical

rese

ction,

nodal

metas

tasis,

tumor

grow

thpattern,

lympho

vasc

ular

inva

sion

,lympho

cytic

infiltration,

DNA

aneu

ploidystatus

,and

synthe

sispha

sefrac

tion

status

.

Cod

on13

,1.93

(1.17–

3.18

)

19Rothan

dco

lleag

ues(201

0)Man

y1,29

9Colon

II–III

372(–)

102(–)

No

Relap

se-freesu

rvival

Adjusted

bytrea

tmen

tarm

and

stag

e.c.34

G>A

,0.99(0.46–

2.09

)c.34

G>T

,1.40(0.89–

2.21

)c.35

G>A

,0.98(0.72–

1.34

)c.35

G>C

,0.97(0.48–

1.95

)c.35

G>T

,1.09(0.76–

1.57

)c.38

G>A

,0.99(0.68–

1.44

)20

Zlobec

and

colleag

ues(201

0)2

392

Colon

and

rectum

I–III

71(–)

27(–)

No

Can

cer-sp

ecificsu

rvival

Cov

ariatesweretumor

dep

th,

nodal

metas

tasis,

andMSI.

c.35

G>A

,HR¼0.82

(P¼0.04

4)21

Yok

otaan

dco

lleag

ues(201

1)1

229

Colon

and

rectum

Adva

nced

and

recu

rren

ce53

(–)

26(–)

Yes

Univa

riate

HRforov

erall

survival

(vs.

KRASwild

-typ

e/BRAFwild

-typ

eas

areferent;

nomultiv

ariate

HRprovided

)

Multiv

ariate

analysis

includ

edbothBRAFmutan

tsan

dBRAFwild

-typ

etumors.

Cov

ariateswereag

e,se

x,perform

ance

status

,BRAF,

patho

logictype,

number

ofmetas

tasis,

andprese

nceof

metas

tasis(live

r,lung

,and

perito

neum

).

(Con

tinue

don

thefollo

wingpag

e)

Imamura et al.





prognostic roles of various KRAS mutations, independentof BRAF mutation status, we used BRAF wild-type tumorsonly and compared KRAS-mutated BRAF wild-type casesto KRASwild-type/BRAFwild-type cases. Thus, among the1,261 patients with KRAS and BRAF data, 181 BRAF-mutated cases were excluded (Fig. 1). Within the remain-ing 1,080 cases, 5 cases with KRAS mutations in bothcodons 12 and 13 were excluded to analyze the prognos-tic effect of KRAS codon 12 mutation separately from thatof codon 13 mutation. As a result, a total of 1,075 BRAFwild-type cases were used for survival analyses (Fig. 1,Supplementary Table S1). There were 512 deaths, includ-ing 299 colorectal cancer–specific deaths, during a medi-an follow-up of 11.7 years (interquartile range, 8.3–16.1years) for censored cases.

The 5-year colorectal cancer–specific survival probabili-ties were 81.5% for patients with KRAS wild-type/BRAFwild-type, 68.8% for those with KRAS codon 12 mutations(with wild-type BRAF), and 75.3% for those with KRAScodon 13 mutations (with wild-type BRAF; Fig. 2 Panels Aand B). Comparedwith patients withKRASwild-type/BRAFwild-type cancers, those with KRAS codon 12 mutationsexperienced a significant increase in colorectal cancer–spe-cific mortality in Kaplan–Meier analysis (log-rank P ¼0.0001) and in Cox regression analysis [univariate HR,1.68, 95% confidence interval (CI), 1.32–2.14, P <0.0001; multivariate HR, 1.30; 95% CI, 1.02–1.67, P ¼0.037; Table 3]. In contrast, comparedwithKRASwild-type/BRAF wild-type cases, patients with KRAS codon 13 muta-tions did not experience any significant reduction in sur-vival (Table 3).

Among the 7 most common KRAS codon 12 and 13mutations, c.35G>T (p.G12V; N ¼ 93) was associated withsignificantly higher colorectal cancer–specific mortality(log-rank P ¼ 0.0007; multivariate HR, 2.00; 95% CI,1.38–2.90, P ¼ 0.0003) compared with KRAS wild-type/BRAF wild-type (Fig. 2 Panels C–D, Table 4). In addition,c.34G>C (p.G12R; N ¼ 8) was associated with highercolorectal cancer–specific mortality (univariate HR, 4.22;95%CI, 1.72–10.4, P¼ 0.0017;multivariateHR, 3.39; 95%CI, 1.28–9.00, P ¼ 0.014) compared with KRAS wild-type/BRAF wild-type. Nonetheless, due to the lower statisticalpower, as well as multiple hypothesis testing (requiring anadjusted stringent significance level at P¼ 0.05/7¼ 0.007),our findings from mutation-specific survival analysesrequired validation in independent datasets. Analysesexcluding the cases with unknown disease stage (N ¼110), yielded similar results (Supplementary Table S2).

To assess the impact of confounding by BRAF mutation,we repeated the above survival analyses including BRAF-mutated cases, most of which were included in the KRASwild-type group. A total case number for this additionalanalysis was 1,255, including 180 BRAF-mutated cases.Compared with KRAS wild-type cases, KRAS codon 12mutations were not significantly associated with colorectalcancer–specific mortality in multivariate analysis, andthe HR effect estimate was substantially attenuated (Sup-plementary Table S3). Among the 7 most common KRAS

Tab

le1.

Studieson

progn

ostic

sign

ifica

nceof

KRASco

don

12an

d13

mutations

inco

lorectal

canc

er(Con

t'd)

No.o

fKRASmutan

tsMultiva

riateHR

(95%

CI;vs

.KRAS-

Ref.

Autho

rs(y)

No.o

fho

spitals

Sam

ple

size

Tum

or

loca

tion

Disea

sestag

eco

don12

(No.o

fev

ents)

codon13

(No.o

fev

ents)

BRAF

data

wild

-typ

eas

areferent,

unless

otherwisesp

ecified

)Notes

Cod

on12

,1.28(0.74–

2.19

)Cod

on13

,2.03(1.10–

3.74

)Im

amuraan

dco

lleag

ues(current

stud

y)Man

y1,26

1(107

5BRAF-w

ild-

typetumors)

Colon

and

rectum

I–IV

332(119

)10

8(31)

Yes

Can

cer-sp

ecificsu

rvival

(vs.

KRASwild

type/BRAF

wild

typeas

areferent)

Adjusted

bystag

e;co

varia

tes

wereag

e,se

x,ye

arof

diagn

osis,tumor

loca

tion,

tumor

differen

tiatio

n,family

historyof

colorectalca

ncer

inan

yfirstd

egreerelativ

e,MSI,

CpG

island

methy

lator

phe

notype,

PIK3C

A,a

ndLINE-1

methy

latio

n.

Cod

on12

,1.29(1.01–

1.65

)Cod

on13

,0.86(0.58–

1.27

)c.34

G>A

,1.00(0.42–

2.34

)c.34

G>C

,3.21(1.22–

8.45

)c.34

G>T

,1.52(0.90–

2.58

)c.35

G>A

,1.09(0.78–

1.53

)c.35

G>C

,0.55(0.24–

1.28

)c.35

G>T

,1.94(1.34–

2.80

)c.38

G>A

,0.88(0.59–

1.30

)






Table 2. Clinical, pathologic, and molecular characteristics according to KRAS mutation status

KRAS mutant

P (wild type

Mutation in either codon 12or codon 13 only

Mutations inboth codon 12and codon 13

Clinical, pathologic, TotalKRASwild type

vs. allmutants Codon 12 P (codon 12 Codon 13

or molecular feature No. (%) No. (%) together) No. (%) vs. codon 13) No. (%) No. (%)

Total no. of patients 1,261 810 335 110 6Sex 0.0049 0.39Male 568 (45%) 341 (42%) 164 (49%) 59 (54%) 4 (67%)Female 693 (55%) 469 (58%) 171 (51%) 51 (46%) 2 (33%)

Mean age (y) � SD 68.5 � 8.7 68.2 � 8.7 0.18 69.4 � 8.6 0.044 67.4 � 9.1 68.7 � 5.2Year of diagnosis 0.48 0.0079Before 1995 396 (31%) 246 (30%) 105 (31%) 41 (37%) 4 (67%)1995 –1999 405 (32%) 260 (32%) 100 (30%) 44 (40%) 1 (17%)2000 –2006 460 (36%) 304 (38%) 130 (39%) 25 (23%) 1 (17%)

Family history ofcolorectalcancer in firstdegree relative(s)

0.35 0.95

Absent 1,020 (81%) 649 (80%) 275 (82%) 90 (82%) 6 (100%)Present 241 (19%) 161 (20%) 60 (18%) 20 (18%) 0

Tumor location

codon 12 and 13 mutations, compared with KRAS wild-type cases, the HR effect estimates for c.34G>C(p.G12R) and c.35G>T (p.G12V) mutations were consid-erably attenuated.

KRAS mutation status and mortality in strata of othervariablesAs exploratory analyses, we examined the prognostic

association of KRAS codon 12 and 13 mutation statusamong BRAF wild-type tumors in various strata, includingdisease stage, sex, age, family history of colorectal cancer,tumor location, tumor differentiation, MSI, CIMP, PIK3CA,and LINE-1 methylation. We did not observe considerableor significant modifying effect by any of these variables onKRAS codon 12 or 13 mutation [all Pinteraction > 0.02; givenmultiple hypothesis testing, a statistical significance levelwas adjusted to Pinteraction ¼ 0.0021)].DiscussionWe conducted this study to assess whether KRAS codon

12–mutated tumors represent a more aggressive subtype ascompared with either KRAS codon 13–mutated tumors orKRAS wild-type tumors, within a group of BRAF wild-typetumors (i.e., controlling for BRAF mutation status). Weshowed that KRAS codon 12 mutations, but not codon13 mutations, were associated with significantly highermortality compared with KRAS wild-type/BRAF wild-typecases. In particular, c.35G>T (p.G12V) mutation was asso-ciated with the highest colorectal cancer–specific mortality(multivariate HR, 2.00; 95% CI, 1.38–2.90, P ¼ 0.0003).

Our data are consistent with previous laboratory studies(24, 25), suggesting that the presence of amutation inKRAScodon 12 confers substantially greater oncogenic potentialas compared with codon 13 mutation. Our data are alsoconsistent with a recent study that showed thatKRAS codon12 mutations, but not codon 13 mutations, conferredresistance to cetuximab in advanced colorectal cancer (15).

Detection of somatic molecular aberrations and tumormolecular classification are increasingly important in colo-rectal cancer (36–39).We used pyrosequencing technology,which has been shown to be more sensitive than Sangersequencing in detecting KRAS mutations in paraffin-embedded archival tissue (30, 40, 41). Pyrosequencing isa sensitive sequencing assay and can reliably detect mutantalleles of low abundance (10% mutant) among wild-typealleles, which is a common situation in solid tumors(30, 40, 41).

To the best of our knowledge based on the literaturesearch in Pubmed, this is the first study to address theprognostic difference between KRAS codon 12 and codon13 mutations in more than 1,000 of BRAF wild-type colo-rectal cancers (i.e., controlling for BRAF mutation status).Although several previous studies have distinguishedbetween the prognostic associations of KRAS mutations incodon 12 and codon 13 (Table 1; refs. 16–21), none of thelarge studies (with a sample size N � 300; refs. 16, 17, 19,20) controlled for BRAF mutation status in their analyses,and the results are conflicting. Given the consistent signif-icant negative prognostic impact of BRAF mutations onpatient survival (10, 11, 19, 21–23), and its associationwithKRAS wild-type tumors, the presence of patients harboringBRAF-mutated tumors within a KRAS wild-type controlgroup would attenuate any negative prognostic effect asso-ciated with KRAS mutation status. Therefore, BRAF muta-tion status must be controlled to assess the precise onco-genic effect of KRAS mutation status. A simple way ofcontrolling for BRAF mutation status is to examine theprognostic role of KRAS mutation in BRAF wild-typetumors. Indeed, BRAF mutation confounded and attenu-ated the negative prognostic effects of KRAS mutations inthis study.

Regulation of RAS involves binding of GTP, which acti-vates the protein. Activation of RAS enables high affinityinteractions with downstream effectors such as RAF-MAPKand phosphoinositide 3-kinase. Subsequently, slow intrin-sicGTPase activity leads to RAS functional inactivation. Thison and off switch regulation is tightly controlled by ARH-GAP (Rho-GTPase activating proteins) and RAPGEF (Rapguanine-nucleotide exchange factors; ref. 42). Interestingly,RAS mutants are resistant to ARHGAP-mediated GTPaseactivation, leading to elevated cellular levels of RAS-GTP(42). Guerrero and colleagues (24) found that KRAS codon12 mutation, by altering the threshold for induction ofapoptosis, confers amore aggressive tumor phenotype thancodon 13 mutation. This suggests that codon 12 mutationresults in greater resistance to ARHGAP-mediated GTPaseactivation than codon 13 mutation (24). Consequently,codon 12–mutated RAS theoretically remains in an active

Colorectal cancer cases withKRAS and BRAF data available

N = 1,261

BRAF mutantN = 181

BRAF wild typeN = 1,080

KRAS mutations in bothcodons 12 and 13

N = 5

KRAS mutations in eithercodons 12 and 13, or wild type

N = 1,075

KRAS wild typeBRAF wild type

N = 635

KRAS codon 12mutant

BRAF wild typeN = 332

KRAS codon 13mutant

BRAF wild typeN = 108

Figure 1. Flow chart of this study. BRAF-mutated cases (N ¼ 181) wereexcluded from survival analysis to assess a prognostic role of KRASmutation in BRAF-wild-type tumors. In addition, cases with KRASmutations in both codons 12 and 13 (N ¼ 5) were excluded, to assess aprognostic effect of KRAS codon 12 mutations separately from that ofKRAS codon 13 mutations.






GTP-bound state longer than codon 13–mutated or wild-type RAS. Experimental data suggest that, among the manydifferent KRAS codon 12 mutations, c.34G>C (p.G12R)and c.35G>T (p.G12V) mutations confer more potenttransforming ability than other KRASmutations, includingc.34G>A (p.G12S), c.34G>T (p.G12C), c.35G>A (p.G12D),and c.35G>C (p.G12A; ref. 43). Moreover, the GTPaseactivity of c.34G>C (p.G12R) and c.35G>T (p.G12V)mutants is lower than that of other KRAS mutations(25, 44). These experimental data are consistent with ourobservations that KRAS codon 12 mutation, especiallyc.34G>C (p.G12R) and c.35G>T (p.G12V), might be asso-ciated with more aggressive tumor behavior. Our findings

underscore the importance of our awareness that differentmutations (even in a single gene) may contribute to differ-ent tumor characteristics and support the unique tumorprinciple (45–47).

Limitations of this study include the lack of data oncancer treatment. Chemotherapeutic and surgical interven-tions have a significant impact on disease progression inmetastatic colorectal cancer. We cannot exclude the possi-bility that there may have been an imbalance in the use oftherapeutic interventions between subgroups. KRAS muta-tion status has recently become an important biomarkerwhen selecting chemotherapeutic agents for colorectal can-cer therapy (6–8, 48). Given that we could not control for

1.0

0.8

0.6

0.4

0.2

0.00 2 4 6 8 10

Colorectal cancer–specific survival (y)

Colorectal cancer(BRAF wild type)

A B

C D

KRAS wild type

KRAS codon 13 mutant


KRAS wild type



KRAS-wild-type

KRAS c.35G>T (p.G12V)

KRAS c.34G>C (p.G12R)

KRAS wild type

KRAS c.35G>T (p.G12V)

KRAS c.34G>C (p.G12R)



Overall log-rankP = 0.0001

Overall log-rankP < 0.0001

Overall log-rankP < 0.0003

Overall log-rankP = 0.0003

Log-rank P = 0.001

Log-rank P = 0.0007


Overall survival (y)

Overall survival (y)Colorectal cancer–specific survival (y)

Sur

viva

l pro

babi

lity

1.0

0.8

0.6

0.4

0.2

0.00 2 4 6 8 10

Sur

viva

l pro

babi

lity

1.0

0.8

0.6

0.4

0.2

0.0

0 2 4 6 8 10

Sur

viva

l pro

babi

lity

1.0

0.8

0.6

0.4

0.2

0.0

0 2 4 6 8 10

Sur

viva

l pro

babi

lity

Figure 2. Kaplan–Meier curves ofBRAF wild-type colorectal cancerpatients according to KRASmutation status. KRAS-mutatedBRAF wild-type cases werecompared with KRAS wild-type/BRAF wild-type cases to assess aprognostic role of KRAS mutationindependent of BRAF mutationstatus. Table indicates the numberof patients who were alive and atrisk of death at each timepoint afterdiagnosis of colorectal cancer. A,colorectal cancer–specific survivalaccording to KRAS codon 12 or 13mutation status. B, overall survivalaccording to KRAS codon 12 or 13mutation status. C, colorectalcancer–specific survival accordingto KRAS c.35G>T (p.G12V) orc.34G>C (p.G12R)mutation status.D, overall survival according toKRAS c.35G>T (p.G12V) orc.34G>C (p.G12R)mutation status.

Imamura et al.





use of EGFR inhibitors, such as cetuximab and panitumu-mab, bias may have arisen through selective use of theseagentswithin the study group.Nonetheless, it seems unlike-ly that chemotherapy use or regimen differed substantiallyby tumor KRAS mutation status, as a vast majority of cases

were diagnosed in 1990s to early 2000s, before 2006, whenKRAS mutation emerged as a predictive biomarker in stageIV colorectal cancer. In addition, our molecular data werenot available for patients or clinicians for treatment deci-sion-making. Another weakness of this study is the absence

Table 3. Colorectal cancer patient mortality according to KRAS mutation status in 1075 BRAF wild typecases

Colorectal cancer–specific mortality Overall mortality

KRAS BRAF Total NNo. ofevents

UnivariateHR (95% CI)

MultivariateHR (95% CI)

No. ofevents



Wild type Wild type 635 149 1 (referent) 1 (referent) 275 1 (referent) 1 (referent)All codon 12mutants

Wild type 332 119 1.68 (1.32–2.14) 1.30 (1.02–1.67) 183 1.46 (1.21–1.77) 1.24 (1.02–1.51)

P < 0.0001 P ¼ 0.037 P < 0.0001 P ¼ 0.029All codon 13mutants

Wild type 108 31 1.25 (0.85–1.84) 0.86 (0.58–1.27) 54 1.17 (0.87–1.57) 0.96 (0.71–1.30)

NS NS NS NS

NOTE: We tested the specific study hypothesis on the prognostic role of KRAS codon 12 mutations, among BRAF wild type cases.Thus, a P value for significance was set at P ¼ 0.05.The multivariate, stage-stratified Cox regression model initially included age, sex, year of diagnosis, tumor location, tumor differen-tiation, family history of colorectal cancer, MSI, CpG island methylator phenotype, PIK3CA, and LINE-1 methylation. A backwardstepwise elimination with a threshold of P ¼ 0.20 was used to select variables in the final model.Abbreviation: NS, not significant.

Table 4. Colorectal cancer patient mortality among the 7 most common KRAS codon 12–13 mutations in1075 BRAF wild-type cases

Colorectal cancer–specific mortality Overall mortality

KRAS BRAF Total NNo. ofevents



No. ofevents



Wild type Wild type 635 149 1 (referent) 1 (referent) 275 1 (referent) 1 (referent)c.34G>A (p.G12S) Wild type 12 6 2.61 (1.15–5.91) 1.03 (0.44–2.44) 7 1.63 (0.77–3.47) 0.96 (0.44–2.08)

P ¼ 0.022 NS NS NSc.34G>C (p.G12R) Wild type 8 5 4.22 (1.72–10.4) 3.39 (1.28–9.00) 6 2.79 (1.24–6.31) 3.19 (1.32–7.69)

P ¼ 0.0017 P ¼ 0.014 P ¼ 0.014 P ¼ 0.010c.34G>T (p.G12C) Wild type 44 16 1.70 (1.01–2.85) 1.56 (0.92–2.65) 25 1.48 (0.98–2.24) 1.42 (0.93–2.17)

P ¼ 0.045 NS NS NSc.35G>A (p.G12D) Wild type 155 49 1.47 (1.07–2.04) 1.08 (0.76–1.51) 79 1.38 (1.07–1.77) 1.16 (0.89–1.51)

P ¼ 0.019 NS P ¼ 0.013 NSc.35G>C (p.G12A) Wild type 19 6 1.36 (0.60–3.08) 0.56 (0.24–1.30) 9 1.03 (0.53–2.01) 0.60 (0.31–1.19)

NS NS NS NSc.35G>T (p.G12V) Wild type 93 37 1.84 (1.28–2.64) 2.00 (1.38–2.90) 57 1.61 (1.21–2.14) 1.54 (1.15–2.07)

P ¼ 0.0010 P ¼ 0.0003 P ¼ 0.0012 P ¼ 0.0042c.38G>A (p.G13D) Wild type 103 31 1.33 (0.90–1.96) 0.88 (0.59–1.30) 52 1.21 (0.90–1.62) 0.98 (0.72–1.35)

NS NS NS NS

NOTE: Themultivariate Cox regressionmodel included the same set of covariates selected as in Table 3. AP value for significancewasadjusted for multiple hypothesis testing to P ¼ 0.05/7 ¼ 0.007. Thus, a P value between 0.05 and 0.007 should be regarded as ofborderline significance.Abbreviation: NS, not significant.






of data on cancer recurrence, and, as a result, disease-freesurvival was not an available outcome measurement inthese cohorts. Because the median survival of metastaticcolorectal cancer patients was 10 to 12 months during thetime period of this study (49), we believe that colorectalcancer–specific survival is a reasonably robust surrogate forcancer-specific outcomes. In fact, disease-free survival hasbeen shown to be highly correlated with overall survival(50).

Strengths of this study include the use of data from 2U.S.nationwide prospective cohort studies. Information ondisease staging, family history of cancer, and other clinico-pathologic and tumor molecular data was prospectivelyintegrated into the molecular pathologic epidemiologydatabase (26–28). Cohort participants whowere diagnosedwith colorectal cancer were presented and treated at hospi-tals throughout the United States, and thus more represen-tative colorectal cancers in the general U.S. population thanare patients in single or few academic medical centers.Finally, by virtue of ourmolecular pathologic epidemiology(26–28) database, we assessed the effects of KRAS codon 12and 13mutations independent of various clinicopathologicfeatures and other critical molecular events such as BRAFand PIK3CA mutations, MSI, CIMP, and LINE-1 hypo-methylation, all of which have been associated with colo-rectal cancer prognosis (2, 34).

In conclusion, our study of more than 1,000 colorectalcancers has shown that KRAS codon 12 mutation (inparticular, c.35G>T, p.G12V), but not codon 13 muta-tion, is associated with worse prognosis in BRAF wild-type colorectal cancers. Different mutations in a singlegene may have distinct biologic effects and clinical impli-cations (47). Because, controlling for BRAF status, KRAScodon 12 mutations contribute to poor prognosis incolorectal cancer, it might be prudent to control for

mutations in BRAF and KRAS in the study arms of futureclinical trials.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed. The content is solely the

responsibility of the authors and does not necessarily represent the officialviews of NCI or NIH.

Authors' ContributionsConception and design: Y. Imamura, K.M. Haigis, C.S. Fuchs, S. OginoDevelopment of methodology: C.S. Fuchs, S. OginoAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.): Y. Imamura, T. Morikawa, X. Liao, M. Yamauchi,Z.R. Qian, C.S. Fuchs, S. OginoAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): Y. Imamura, T. Morikawa, P. Lochhead,A. Kuchiba, J.A. Meyerhardt, C.S. Fuchs, S. OginoWriting, review, and/or revision of the manuscript: Y. Imamura, T.Morikawa, X. Liao, P. Lochhead, J.A. Meyerhardt, C.S. Fuchs, S. OginoAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): T. Morikawa, Z.R. Qian, R. Nishi-hara, C.S. Fuchs, S. OginoStudy supervision: C.S. Fuchs, S. Ogino

AcknowledgmentsThe authors thank hospitals and pathology departments throughout the

United States for generously providing us with tissue specimens. In addition,the authors thank the participants and staff of the Nurses’ Health Study andtheHealth Professionals Follow-UpStudy, for their valuable contributions aswell as the U.S. state cancer registries for their help.

Grant SupportThis work was supported by the NIH [P01 CA87969 (to S.E. Hankinson),

P01 CA55075 (to W.C. Willett), P50 CA127003 (to C.S. Fuchs), and R01CA151993 (to S. Ogino)]; the Bennett Family Fund for Targeted TherapiesResearch and the Entertainment Industry Foundation through NationalColorectal Cancer Research Alliance. T. Morikawa was supported by a fel-lowship grant from the Japan Society for Promotion of Science. P. Lochheadwas supported by Frank KnoxMemorial Fellowship fromHarvardUniversity.

The costs of publication of this articlewere defrayed in part by thepaymentof page charges. This articlemust therefore be herebymarked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received December 13, 2011; revised June 8, 2012; accepted June 22,2012; published OnlineFirst July 2, 2012.

References1. Janakiraman M, Vakiani E, Zeng Z, Pratilas CA, Taylor BS, Chitale D,

et al. Genomic and biological characterization of exon 4 KRAS muta-tions in human cancer. Cancer Res 2010;70:5901–11.

2. Nosho K, Irahara N, Shima K, Kure S, Kirkner GJ, Schernhammer ES,et al. Comprehensive biostatistical analysis of CpG island methylatorphenotype in colorectal cancer using a large population-based sam-ple. PLoS One 2008;3:e3698.

3. Roock WD, Vriendt VD, Normanno N, Ciardiello F, Tejpar S. KRAS,BRAF, PIK3CA, and PTEN mutations: implications for targeted ther-apies in metastatic colorectal cancer. Lancet Oncol 2011;12:594–603.

4. Allegra CJ, Jessup JM, Somerfield MR, Hamilton SR, Hammond EH,Hayes DF, et al. American Society of Clinical Oncology provisionalclinical opinion: testing for KRAS gene mutations in patients withmetastatic colorectal carcinoma to predict response to anti-epidermalgrowth factor receptor monoclonal antibody therapy. J Clin Oncol2009;27:2091–6.

5. Jimeno A, Messersmith WA, Hirsch FR, Franklin WA, Eckhardt SG.KRAS mutations and sensitivity to epidermal growth factor receptorinhibitors in colorectal cancer: practical application of patient selec-tion. J Clin Oncol 2009;27:1130–6.

6. Van Cutsem E, Kohne CH, Lang I, Folprecht G, Nowacki MP, CascinuS, et al. Cetuximab plus irinotecan, fluorouracil, and leucovorin as first-line treatment for metastatic colorectal cancer: updated analysis of

overall survival according to tumor KRAS andBRAFmutation status. JClin Oncol 2011;29:2011–9.

7. Douillard JY, Siena S, Cassidy J, Tabernero J, Burkes R, Barugel M,et al. Randomized, phase III trial of panitumumab with infusionalfluorouracil, leucovorin, and oxaliplatin (FOLFOX4) versus FOLFOX4alone as first-line treatment in patients with previously untreatedmetastatic colorectal cancer: the PRIME study. J Clin Oncol2010;28:4697–705.

8. Bokemeyer C, Bondarenko I, Makhson A, Hartmann JT, Aparicio J, deBraud F, et al. Fluorouracil, leucovorin, and oxaliplatin with andwithoutcetuximab in the first-line treatment of metastatic colorectal cancer. JClin Oncol 2009;27:663–71.

9. Zlobec I, Bihl MP, Schwarb H, Terracciano L, Lugli A. Clinicopatho-logical and protein characterization of BRAF- and K-RAS-mutatedcolorectal cancer and implications for prognosis. Int J Cancer2010;127:367–80.

10. Farina-Sarasqueta A, van Lijnschoten G, Moerland E, Creemers GJ,Lemmens VE, Rutten HJ, et al. The BRAF V600E mutation is anindependent prognostic factor for survival in stage II and stage IIIcolon cancer patients. Ann Oncol 2010;21:2396–402.

11. Richman SD, SeymourMT, Chambers P, Elliott F, Daly CL,MeadeAM,et al. KRAS and BRAF mutations in advanced colorectal cancer areassociated with poor prognosis but do not preclude benefit from

Imamura et al.





oxaliplatin or irinotecan: results from the MRC FOCUS trial. J ClinOncol 2009;27:5931–7.

12. Samowitz WS, Curtin K, Wolff RK, Tripp SR, Caan BJ, Slattery ML.Microsatellite instability and survival in rectal cancer. Cancer CausesControl 2009;20:1763–8.

13. Barault L, Charon-Barra C, Jooste V, de la Vega MF, Martin L, RoignotP, et al. Hypermethylator phenotype in sporadic colon cancer: studyona population-based series of 582 cases. Cancer Res 2008;68:8541–6.

14. Ogino S, Meyerhardt JA, Irahara N, Niedzwiecki D, Hollis D, Saltz LB,et al. KRAS mutation in stage III colon cancer and clinical outcomefollowing intergroup trial CALGB 89803. Clin Cancer Res 2009;15:7322–9.

15. De Roock W, Jonker DJ, Di Nicolantonio F, Sartore-Bianchi A, Tu D,Siena S, et al. Association of KRAS p.G13D mutation with outcome inpatients with chemotherapy-refractory metastatic colorectal cancertreated with cetuximab. JAMA 2010;304:1812–20.

16. Samowitz WS, Curtin K, Schaffer D, Robertson M, Leppert M, SlatteryML. Relationship of Ki-ras mutations in colon cancers to tumor loca-tion, stage, and survival: a population-based study. Cancer EpidemiolBiomarkers Prev 2000;9:1193–7.

17. Andreyev HJ, Norman AR, Cunningham D, Oates J, Dix BR, IacopettaBJ, et al. Kirsten ras mutations in patients with colorectal cancer: the'RASCAL II' study. Br J Cancer 2001;85:692–6.

18. Bazan V,MigliavaccaM, Zanna I, Tubiolo C, Grassi N, Latteri MA, et al.Specific codon 13K-rasmutations are predictive of clinical outcome incolorectal cancer patients, whereas codon 12 K-ras mutations areassociated with mucinous histotype. Ann Oncol 2002;13:1438–46.

19. Roth AD, Tejpar S, Delorenzi M, Yan P, Fiocca R, Klingbiel D, et al.Prognostic role of KRAS and BRAF in stage II and III resected coloncancer: results of the translational study on the PETACC-3, EORTC40993, SAKK 60-00 trial. J Clin Oncol 2010;28:466–74.

20. Zlobec I, Kovac M, Erzberger P, Molinari F, Bihl MP, Rufle A, et al.Combined analysis of specific KRAS mutation, BRAF and microsat-ellite instability identifies prognostic subgroups of sporadic and hered-itary colorectal cancer. Int J Cancer 2010;127:2569–75.

21. Yokota T, Ura T, Shibata N, Takahari D, Shitara K, Nomura M, et al.BRAF mutation is a powerful prognostic factor in advanced andrecurrent colorectal cancer. Br J Cancer 2011;104:856–62.

22. SamowitzWS, Sweeney C, Herrick J, Albertsen H, Levin TR,MurtaughMA, et al. Poor survival associated with the BRAF V600E mutation inmicrosatellite-stable colon cancers. Cancer Res 2005;65:6063–9.

23. French AJ, Sargent DJ, Burgart LJ, Foster NR, Kabat BF, Goldberg R,et al. Prognostic significance of defective mismatch repair and BRAFV600E in patients with colon cancer. Clin Cancer Res 2008;14:3408–15.

24. Guerrero S, Casanova I, Farre L,Mazo A, Capella G,Mangues R. K-rascodon 12mutation induces higher level of resistance to apoptosis andpredisposition to anchorage-independent growth than codon 13mutation or proto-oncogene overexpression. Cancer Res 2000;60:6750–6.

25. Bollag G, McCormick F. Intrinsic and GTPase-activating protein-stim-ulated Ras GTPase assays. Methods Enzymol 1995;255:161–70.

26. Ogino S, Chan AT, Fuchs CS, Giovannucci E. Molecular pathologicalepidemiology of colorectal neoplasia: an emerging transdisciplinaryand interdisciplinary field. Gut 2011;60:397–411.

27. Ogino S, Galon J, Fuchs CS, Dranoff G. Cancer immunology–analysisof host and tumor factors for personalized medicine. Nat Rev ClinOncol 2011;8:711–9.

28. Ogino S, Stampfer M. Lifestyle factors and microsatellite instability incolorectal cancer: The evolving field of molecular pathological epide-miology. J Natl Cancer Inst 2010;102:365–367.

29. Morikawa T, Kuchiba A, Yamauchi M, Meyerhardt JA, Shima K, NoshoK, et al. Association of CTNNB1 (beta-catenin) alterations, body massindex, and physical activity with survival in patients with colorectalcancer. JAMA 2011;305:1685–94.

30. Ogino S, Kawasaki T, Brahmandam M, Yan L, Cantor M, Namgyal C,et al. Sensitive sequencing method for KRAS mutation detection bypyrosequencing. J Mol Diagn 2005;7:413–21.

31. Ogino S, Kawasaki T, Kirkner GJ, Loda M, Fuchs CS. CpG islandmethylator phenotype-low (CIMP-low) in colorectal cancer: possible

associations with male sex and KRAS mutations. J Mol Diagn2006;8:582–8.

32. Liao X, Morikawa T, Lochhead P, Imamura Y, Kuchiba A, YamauchiM, et al. Prognostic role of PIK3CA mutation in colorectal cancer:cohort study and literature review. Clin Cancer Res 2012;18:2257–68.

33. WeisenbergerDJ,SiegmundKD,CampanM,YoungJ, LongTI, FaasseMA, et al. CpG island methylator phenotype underlies sporadic micro-satellite instability and is tightly associated with BRAF mutation incolorectal cancer. Nat Genet 2006;38:787–93.

34. Ogino S, Nosho K, Kirkner GJ, Kawasaki T, Chan AT, SchernhammerES, et al. A cohort study of tumoral LINE-1 hypomethylation andprognosis in colon cancer. J Natl Cancer Inst 2008;100:1734–8.

35. Irahara N, Nosho K, Baba Y, Shima K, Lindeman NI, Hazra A, et al.Precision of pyrosequencing assay to measure LINE-1 methylation incolon cancer, normal colonicmucosa, andperipheral blood cells. JMolDiagn 2010;12:177–83.

36. TanakaM,ChangP, Li Y, Li D, OvermanM,MaruDM, et al. Associationof CHFR promoter methylation with disease recurrence in locallyadvanced colon cancer. Clin Cancer Res 2011;17:4531–40.

37. Dahlin AM, Palmqvist R, Henriksson ML, Jacobsson M, Eklof V,Rutegard J, et al. The role of the CpG island methylator phenotypein colorectal cancer prognosis depends on microsatellite instabilityscreening status. Clin Cancer Res 2010;16:1845–55.

38. Yagi K, Akagi K, Hayashi H, Nagae G, Tsuji S, Isagawa T, et al. ThreeDNA methylation epigenotypes in human colorectal cancer. Clin Can-cer Res 2010;16:21–33.

39. Balaguer F, Moreira L, Lozano JJ, Link A, Ramirez G, Shen Y, et al.Colorectal cancerswithmicrosatellite instability display uniquemiRNAprofiles. Clin Cancer Res 2011;17:6239–49.

40. Weichert W, Schewe C, Lehmann A, Sers C, Denkert C, Budczies J,et al. KRAS genotyping of paraffin-embedded colorectal cancer tissuein routine diagnostics: comparison of methods and impact of histol-ogy. J Mol Diagn 2010;12:35–42.

41. Tsiatis AC, Norris-Kirby A, Rich RG, Hafez MJ, Gocke CD, EshlemanJR, et al. Comparison of Sanger sequencing, pyrosequencing, andmelting curve analysis for the detection of KRASmutations: diagnosticand clinical implications. J Mol Diagn 2010;12:425–32.

42. Karnoub AE, Weinberg RA. Ras oncogenes: split personalities. NatRev Mol Cell Biol 2008;9:517–31.

43. Seeburg PH, Colby WW, Capon DJ, Goeddel DV, Levinson AD.Biological properties of human c-Ha-ras1 genes mutated at codon12. Nature 1984;312:71–5.

44. Al-Mulla F,Milner-White EJ,Going JJ, BirnieGD.Structural differencesbetween valine-12 and aspartate-12 Ras proteins may modify carci-noma aggression. J Pathol 1999;187:433–8.

45. Ogino S, Goel A. Molecular classification and correlates in colorectalcancer. J Mol Diagn 2008;10:13–27.

46. Ogino S, Giovannucci E. Lifestyle factors and colorectal cancer micro-satellite instability: molecular pathological epidemiology science,based on unique tumour principle. In J Epidemiol 2012. May 16. [Epubahead of print].

47. Ogino S, Fuchs CS, Giovannucci E. How many molecular subtypes?Implications of the unique tumor principle in personalized medicine.Expert Rev Mol Diagn 2012;12:621–8.

48. Moosmann N, von Weikersthal LF, Vehling-Kaiser U, Stauch M, HassHG, Dietzfelbinger H, et al. Cetuximab plus capecitabine and irinote-can compared with cetuximab plus capecitabine and oxaliplatin asfirst-line treatment for patients with metastatic colorectal cancer: AIOKRK-0104–a randomized trial of the German AIO CRC study group. JClin Oncol 2011;29:1050–8.

49. Ogino S, Nosho K, Irahara N, Shima K, Baba Y, Kirkner GJ, et al.Prognostic significance and molecular associations of 18q loss ofheterozygosity: a cohort study of microsatellite stable colorectal can-cers. J Clin Oncol 2009;27:4591–8.

50. Sargent DJ, Wieand HS, Haller DG, Gray R, Benedetti JK, Buyse M,et al. Disease-free survival versus overall survival as a primary endpoint for adjuvant colon cancer studies: individual patient data from20,898 patients on 18 randomized trials. J Clin Oncol 2005;23:8664–70.






2012;18:4753-4763. Published OnlineFirst July 2, 2012.Clin Cancer Res Yu Imamura, Teppei Morikawa, Xiaoyun Liao, et al.

Wild-Type Colorectal CancersBRAFPrognosis in 1075 Codons 12 and 13, and PatientKRASSpecific Mutations in

Updated version

10.1158/1078-0432.CCR-11-3210doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2012/07/02/1078-0432.CCR-11-3210.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/18/17/4753.full#ref-list-1

This article cites 49 articles, 23 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/18/17/4753.full#related-urls

This article has been cited by 10 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/18/17/4753To request permission to re-use all or part of this article, use this link


http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-11-3210http://clincancerres.aacrjournals.org/content/suppl/2012/07/02/1078-0432.CCR-11-3210.DC1http://clincancerres.aacrjournals.org/content/18/17/4753.full#ref-list-1http://clincancerres.aacrjournals.org/content/18/17/4753.full#related-urlshttp://clincancerres.aacrjournals.org/cgi/alertsmailto:[email protected]://clincancerres.aacrjournals.org/content/18/17/4753http://clincancerres.aacrjournals.org/

speciﬁc mutations in kras codons 12 and 13, and patient ...within the group of kras wild-type...

Documents