cimp status of interval colon cancers: another piece to the puzzle

nature publishing group ORIGINAL CONTRIBUTIONS

CO

LON

/SM

ALL

BO

WE

L

1189

© 2010 by the American College of Gastroenterology The American Journal of GASTROENTEROLOGY

INTRODUCTION Colonoscopy is considered the gold standard test for the detec-

tion and prevention of colon cancer ( 1 ). Tumors diagnosed in

the interval between serial colonoscopy (interval cancers) have a

negative impact on our ability to reduce the incidence of colon

cancer ( 2,3 ). Robertson DJ et al. ( 4 ) reported 58 interval cancers

among 9,000 individuals with one or more adenomas at baseline

over approximately 4 years of follow-up. Some interval cancers

CIMP Status of Interval Colon Cancers: Another Piece to the Puzzle Mustafa A. Arain , MD 1 , Mandeep Sawhney , MD 2 , Shehla Sheikh , MD 1 , Ruth Anway , BA 1 , Bharat Thyagarajan , MD, PhD 3 , John H. Bond , MD 1 and Aasma Shaukat , MD, MPH 1 , 4

OBJECTIVES: Colon cancers diagnosed in the interval after a complete colonoscopy may occur due to limitations of colonoscopy or due to the development of new tumors, possibly refl ecting molecular and environmental differences in tumorigenesis resulting in rapid tumor growth. In a previous study from our group, interval cancers (colon cancers diagnosed within 5 years of a complete colonoscopy) were almost four times more likely to demonstrate microsatellite instability (MSI) than non-interval cancers. In this study we extended our molecular analysis to compare the CpG island methylator phenotype (CIMP) status of interval and non-interval colorectal cancers and investigate the relationship between the CIMP and MSI pathways in the pathogenesis of interval cancers.

METHODS: We searched our institution ’ s cancer registry for interval cancers, defi ned as colon cancers that developed within 5 years of a complete colonoscopy. These were frequency matched in a 1:2 ratio by age and sex to patients with non-interval cancers (defi ned as colon cancers diagnosed on a patient ’ s fi rst recorded colonoscopy). Archived cancer specimens for all subjects were retrieved and tested for CIMP gene markers. The MSI status of subjects identifi ed between 1989 and 2004 was known from our previous study. Tissue specimens of newly identifi ed cases and controls (between 2005 and 2006) were tested for MSI.

RESULTS: There were 1,323 cases of colon cancer diagnosed over the 17-year study period, of which 63 were identifi ed as having interval cancer and matched to 131 subjects with non-interval cancer. Study subjects were almost all Caucasian men. CIMP was present in 57 % of interval cancers compared to 33 % of non-interval cancers ( P = 0.004). As shown previously, interval cancers were more likely than non-interval cancers to occur in the proximal colon (63 % vs. 39 % ; P = 0.002), and have MSI 29 % vs. 11 % , P = 0.004). In multivariable logistic regression model, proximal location (odds ratio (OR) 1.85; 95 % confi dence interval (CI) 1.01 – 3.8), MSI (OR 2.7; 95 % CI 1.1 – 6.8) and CIMP (OR 2.41; 95 % CI 1.2 – 4.9) were independently associated with interval cancers. CIMP was associated with interval cancers independent of MSI status. There was no difference in 5-year survival between the two groups.

CONCLUSIONS: Interval cancers are more likely to arise in the proximal colon and demonstrate CIMP, which suggests there may be differences in biology between these and non-interval CRC. Additional studies are needed to determine whether interval cancers arise as a result of missed lesions or accelerated neoplastic progression.

Am J Gastroenterol 2010; 105:1189–1195; doi: 10.1038/ajg.2009.699; published online 15 December 2009

1 Division of Gastroenterology, Department of Medicine, University of Minnesota , Minneapolis , Minnesota , USA ; 2 Division of Gastroenterology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School , Boston , Massachusetts , USA ; 3 Department of Laboratory Medicine and Pathology, University of Minnesota , Minneapolis , Minnesota , USA ; 4 VA Medical Center , Minneapolis , Minnesota , USA . Correspondence: Aasma Shaukat, MD, MPH , Section of Gastroenterology, Department of Medicine , One Veterans Drive, 111-D, Minneapolis , Minnesota 55417 , USA . E-mail: [email protected] Presented in abstract form at the ACG National Meeting, Orlando October 2008. Received 9 February 2009; accepted 11 November 2009

The American Journal of GASTROENTEROLOGY VOLUME 105 | MAY 2010 www.amjgastro.com

1190 C

OLO

N/S

MA

LL B

OW

EL

Arain et al.

likely occur due to lesions being missed at the time of colonos-

copy or inadequate removal of precursor lesions refl ecting limi-

tations of colonoscopy and colonoscopic technique. It is likely

however that some interval tumors develop over a short period of

time, refl ecting rapid tumor growth. Th e rate of tumor growth is

likely dependant on the underlying molecular pathways involved

in tumor development ( 5 ), however this has not been proven. Th e

comparison of interval cancers with non-interval cancers off ers

a unique means to identify the genetic pathways responsible for

these lesions and may shed light on the mechanisms involved in

rapid tumor growth.

Th ree molecular pathways have been described in the pathogen-

esis of colon cancers — chromosomal instability (CIN), microsatel-

lite instability (MSI) and CpG island methylator phenotype (CIMP)

( 6 – 8 ). Chromosomal instability refers to aneuploidy and loss of

hetrozygosity resulting in progressive accumulation of mutational

changes which lead to loss of tumor suppressor genes (e.g., APC,

DCC / SMAD 4, p53) and / or activation of oncgenes (e.g., K-ras)

as seen in progression of adenomas to carcinomas. Microsatellite

instability is characterized by loss of function of DNA mismatch

repair genes as a result of which mutational changes go uncor-

rected off ering a survival benefi t to tumor cells. Mutations asso-

ciated with CIN and MSI can be hereditary resulting in familial

cancer syndromes or somatic resulting in sporadic cancers ( 8 ).

CIMP is an epigenetic phenomenon characterized by meth-

ylation of cytosine residues of CpG rich islands in the promoter

region of tumor suppressor genes resulting in transcriptional inac-

tivation and therefore loss of expression of these genes. CIMP is

an acquired phenomenon in which the genetic code remains unal-

tered (i.e., aff ected gene loci do not undergo mutational changes)

instead, changes upstream from the gene lead to conformational

changes of chromatin resulting in the genes not being expressed. It

is believed that the precursor lesion for CIMP + colorectal cancers

may be sessile serrated adenomas ( 6 ).

We have previously shown that interval colon cancers are three

times more likely to have microsatellite instability than non-inter-

val cancers ( 9 ). Th e aim of this study was to investigate the CIMP

status of interval and non-interval colorectal cancers and explore

the relationship between the CIMP and MSI pathways in the

pathogenesis of interval cancers.

METHODS All veterans with a diagnosis of cancer at the Minneapolis Veter-

ans Aff airs Medical Center are entered into a cancer registry and

are monitored for lifetime follow-up. Th e registry has consistently

achieved 95 % follow-up. In a previous study, the cancer registry

was searched from January 1, 1989, to December 31, 2004 and all

patients with incident colon cancer were identifi ed. Patients with

infl ammatory bowel disease or familial adenomatous polyposis

were excluded. Patients were defi ned as having interval colon

cancer if they were diagnosed with a primary colon cancer within

5 years of a complete colonoscopy. Colonoscopy was considered

complete if the endoscopist visualized and documented cecal land-

marks and completely removed all visible polyps. Patients were

defi ned as having a non-interval cancer if they were diagnosed

with colon cancer on their fi rst recorded colonoscopy. To account

for age at diagnosis as a confounder, patients with interval cancers

were stratifi ed by median age (older than 70 years or 70 years or

younger). A random sample of patients with non-interval cancer

was chosen from the cancer registry and frequency matched with

patients with interval cancer in a 2:1 ratio for each stratum. At the

time of the current study, cancer registry data for an additional

two years was available. We updated the search through Decem-

ber 2006 and added to our previous study cohorts. In anticipation

that MSI and CIMP analysis may not be feasible in all archived

colon cancer tissue, we oversampled controls by 10 subjects.

CIMP and MSI analysis Archived colon cancer tissue that had been fi xed in formalin and

embedded in paraffi n was retrieved for analysis for all study sub-

jects. Tumor tissue was identifi ed and aft er manual dissection,

deparaffi nized using xylene. Genomic DNA was extracted using

the QIAmp DNA mini kit (Qiagen, Valencia, CA) according to

the manufacturer ’ s instructions.

One microgram of genomic DNA was bisulfi te modifi ed by Epi-

Tect Bisulfi te Kit according to manufacturer ’ s instructions. Real

Time PCR were carried out using 50 ng of bisulfi te converted DNA

in a total volume of 50 μ l, which contained 5 μ l of 10 × reaction

buff er with Magnesium, 0.25 m M of each dNTP, 10 pmol forward

and reverse primers, 100 pmol methylated and unmethylated

probes for fi ve promoters in separate reactions and 2 units of Fast-

start taq Polymerase (Roche). All reactions were run on the Roche

real time PCR machine. We tested for promoter methylation of

the following genes: MINT1, MINT2, MINT31, p16INK4, MGMT,

hMLH1. Th e presence of CIMP (CIMP + ) was defi ned as meth-

ylation of ≥ 3 genes of interest, and absence of CIMP (CIMP − )

defi ned as methylation of < 3 genes. In sensitivity analysis, when

comparing CIMP status with MSI, we removed hMLH1 from the

panel and defi ned CIMP as methylation of promoter of ≥ 2 genes

of interest.

For MSI analysis, samples of genomic DNA were used to amplify

sequences for the following mononucleotide and dinucleotide

microsatellite loci: BAT25, BAT26, BAT34c4, BAT40, D17S250,

actc, D5S346, mycl, D18s55, and D10S197. Using the National

Cancer Institute Workshop on Microsatellite Instability for Cancer

Detection and Familial Predisposition recommendations, tumors

that showed instability in ≥ 40 % of markers tested were designated

as cancers with MSI. Genomic DNA extraction and MSI testing

for DNA samples from the previous study was conducted at Mayo

Clinic Rochester, MN. MSI testing for DNA samples of newly iden-

tifi ed cases for this study was done at the University of Minnesota,

Minneapolis, MN using 5 dinucleotide markers (BAT25, BAT26,

NR21, NR22, NR24). Tumors showing instability in > 40 % of the

markers (2 of 5) were defi ned as MSI tumors.

CIMP analysis could not be conducted for 9 (14 % ) subjects with

interval cancer and 23 (17 % ) subjects with non-interval cancer

and MSI analysis could not be conducted for 8 (12 % ) subjects with

interval cancer and 24 (18 % ) subjects with non-interval cancer.

CIMP and MSI analysis could not be conducted in these subjects


1191

CO

LON

/SM

ALL

BO

WE

L

CIMP Status of Interval Colon Cancers

due to unavailability of tissue specimens or low cellularity of can-

cer tissue. Complete CIMP and MSI data was available for 52

(83 % ) subjects with interval cancer and 103 (78 % ) subjects with

non-interval cancer.

Data collection Data of interest was extracted from patient medical records and

cancer registry case fi les using an abstraction instrument that was

created for the study. Th e AJCC Cancer Staging Manual recom-

mendations were used to defi ne and categorize study variables

such as histologic grade and TNM cancer stage. Th e colon was

divided into 8 segments for describing tumor location and col-

lapsed into 2 categories: proximal colon (cecum, ascending,

hepatic fl exure, and transverse colon), and distal colon (splenic

fl exure, descending colon, sigmoid colon, and rectum). Mucinous

and signet ring carcinoma was defi ned as colon cancer with > 50 %

of the relevant histology. Follow-up and survival information col-

lected for new subjects and updated for previous subjects. Our

institutional review board approved the study.

Statistical analysis Categorical exposure variables were compared using χ 2 test or

Fisher exact test, normally distributed continuous variables using

t -test, and non-normally distributed continuous variables using

Wilcoxon rank sum test. Interval or non-interval colon cancer

status served as the dependent variable. Presence or absence of

MSI and CIMP status served as the primary exposures of inter-

est. Logistic regression was used to adjust for covariates such as

stage of diagnosis, size, histology and location of tumors. Hier-

archical regression was used to determine the fi nal regression

model. Goodness of fi t for regression models was checked using

the Hosmer – Lemeshow χ 2 test. Similar analyses were conducted

using CIMP status and MSI as the dependent variables respec-

tively, to identify factors associated with CIMP and MSI respec-

tively, such as size, location histology of tumors, controlling for

stag at diagnosis. Th e Mcnemar ’ s test for discordance was used

to test the association between CIMP and MSI. Five year sur-

vival for patients with interval cancer and non-interval cancer

was compared using Kaplan – Meier survival analysis. A Wil-

coxon test was used to check for equality of survivor function.

Cox proportional hazards regression models were used to adjust

for age, sex, tumor stage, and cancer treatment. All analysis were

performed using statistical soft ware STATA Intercooled version

10.0 (College Station, TX).

Th is 1:2 case control study had 80 % power to detect diff erences

in CIMP prevalence of 2 times or greater between interval and

non-interval cancers.

RESULTS A total of 1,323 colon cancer cases were identifi ed from the Min-

neapolis Veterans Aff airs Medical Center cancer registry for the

specifi ed study period, and met entry criteria for the study. Sixty

three patients were identifi ed as having an interval cancer (cancer

diagnosed within 5 years of a complete colonoscopy) and were

matched with 131 subjects with non interval cancer (cancer diag-

nosed at the time of index colonoscopy) as described in the meth-

ods section. Mean follow-up for interval and non-interval cancer

patients was 46.4 months (range 2 – 180 months) and 47.1 months

(range 1 – 215 months) respectively. Study subjects were almost all

white (98 % ) and male (98 % ).

A comparison of tumor characteristics between interval and

non-interval subjects is shown in Table 1 . Interval cancers were

more likely than non-interval cancers to occur in the proximal

colon (63 % vs. 39 % ; P = 0.002), be smaller in size at the time of

diagnosis (3.6 cm vs. 4.5 cm; P = 0.001) and have mucinous at his-

tology (32 % vs. 9 % ; P = 0.001). Th ere were no diff erences between

interval and non-interval cases for histologic grade, TNM stage at

diagnosis or treatment with chemotherapy, radiation and surgery,

with respect to stage.

Th e results of CIMP and MSI analysis for interval and non-

interval cancers are displayed in Table 2 . CIMP was demonstrated

in 57 % of the patients with interval cancer compared with 33 % of

patients with non interval cancer ( P = 0.004). Similar to our pre-

vious study, interval cancers were also signifi cantly more likely

to demonstrate MSI than non interval cancers (29 % vs. 11 % ,

P = 0.004). In multivariable logistic regression, proximal location

(OR 1.85; 95 % CI, 1.01 – 3.8), MSI (OR 2.7; 95 % CI 1.10 – 6.8) and

CIMP (OR 2.41; 95 % CI, 1.2 – 4.9) were associated with interval

cancers. Th ere was no interaction by TNM stage of tumor.

Interval and non-interval cancers were pooled and a compari-

son was performed with respect to CIMP ( Table 3 ). CIMP + can-

cers were more likely occur in the proximal colon (58 % vs. 42 % ,

P = 0.02). In sensitivity analysis, we removed hypermethylation

of hMLH1 from panel of CIMP markers, and defi ned CIMP as

Table 1 . Clinicopathologic characteristics of interval and non-interval cancers

Variable Interval ( n = 63) Non-interval

( n =131) P value

Age (years) 75 ± 8 73 ± 8 NS

Follow-up (months) 46.4 ± 5.4 47.1 ± 3.8 NS

Proximal location 40 (63 % ) 51 (39 % ) 0.002

Size (cm) 3.6 ± 2 4.5 ± 2 0.001

Mucinous histology 20 (32 % ) 11 (9 % ) 0.001

Differentiation

Poor 11 (17 % ) 19 (14 % )

Moderate / well 48 (76 % ) 107 (81 % ) NS

Unknown 4 (6 % ) 5 (4 % )

TNM stage

I 19 (31 % ) 33 (26 % )

II 25 (41 % ) 44 (34 % )

III 11 (17 % ) 31 (24 % ) NS

IV 6 (9 % ) 21 (15 % )

Unknown 2 (2 % ) 2 (1 % )

NS, not signifi cant.


1192 C

OLO

N/S

MA

LL B

OW

EL

Arain et al.

methylation of ≥ 2 promoters of interest. Th is did not change the

association of CIMP with interval cancers or with MSI. Similarly, a

pooled analysis was performed with respect to MSI status as well.

Consistent with our previous results, MSI cancers were more likely

than microsatellite stable tumors to occur in the proximal colon

(78 % vs. 43 % , P = 0.001), show mucinous histology (46 % vs. 9 % ;

P = 0.001) and demonstrate poor diff erentiation (46 % vs. 10 % ;

P = 0.001). No such associations were found for CIMP status. Th ere

were signifi cant diff erences in overlap between MSI and CIMP status

of cancers ( P = 0.01), i.e., not all CIMP + cancers demonstrated

MSI + , suggesting more than one pathway may be involved ( Table 4 ).

Th e 5 year survival was not signifi cantly diff erent between inter-

val vs. non-interval cancers (44 % (95 % CI, 27 – 60 % ) for interval

cancers vs. 41 % (95 % , CI, 31 – 52 % ), for non interval cancers,

P = 0.12) ( Figure 1 ). In Cox proportional hazards model, adjust-

ing for age, tumor grade, CIMP and MSI status, and stratifying

by TNM stage we found no statistical diff erence in survival for

interval cancers compared to non-intervals cancers ( P = 0.2). We

also did not fi nd any diff erences in 5 year survival for CIMP + vs.

CIMP − tumors [43 % (95 % CI 29 – 57 % ) for CIMP + vs. 43 % (95 %

Table 2 . Molecular characteristics of interval vs. non-interval cancers

Interval Non-interval P value

CIMP a

Positive 31 (57 % ) 33 (33 % ) 0.004

Negative 23 (43 % ) 75 (66 % )

MSI b

MSI 16 (29 % ) 12 (11 % ) 0.004

MSS 39 (71 % ) 95 (89 % )

CIMP, CpG island methylator phenotype; MSI, microsatellite instability; MSS, microsatellite stable. a n =167. b n =162. All values expressed as number and % with respect to interval / non-interval status.

Table 3 . Characteristics of cancers by CIMP status

CIMP + CIMP − P value

n = 69 ( % ) n = 98 ( % )

Size (cm) 4.0 ± 2.1 4.5 ± 2.3 NS

Location

Proximal 40 (58 % ) 40 (41 % ) 0.02

Distal 29 (42 % ) 58 (57 % )

Mucinous histology 13 (19 % ) 15 (15 % ) NS

Differentiation

Poor 8 (12 % ) 19 (20 % ) NS

Moderate / good 59 (88 % ) 76 (80 % )

MSI a

MSI 13 (20 % ) 13 (15 % ) NS

MSS 52 (80 % ) 77 (85 % )

CIMP, CpG island methylator phenotype; MSI, microsatellite instability; MSS, microsatellite stable; NS, not signifi cant. a n =155.

Table 4 . Characteristics of cancers by MSI status

MSI MSS P value

n =28 ( % ) n =134 ( % )

Size (cm) 4.9 ± 2.5 4.2 ± 2.2 NS

Location

Proximal 21 (78 % ) 57 (42 % ) 0.001

Distal 7 (25 % ) 77 (58 % )

Mucinous histology 13 (46 % ) 13 (9 % ) 0.001

Differentiation

Poor 13 (46 % ) 14 (10 % ) 0.001

Moderate / good 10 (43 % ) 119 (89 % )

CIMP

CIMP + 13 (48 % ) 52 (39 % ) NS

CIMP − 13 (48 % ) 77 (58 % )

Unknown 2 (4 % ) 5 (3 % )

CIMP, CpG island methylator phenotype; MSI, microsatellite instability; MSS, microsatellite stable; NS, not signifi cant.

Figure 1 . Kaplan–Meier curves for 5-year survival for interval and non-interval cancers.

0.00

0.25

0.50

0.75

1.00

Pro

port

ion

surv

ivin

g

0 20 40 60

Analysis time (months)

Non-interval cancers Interval cancers

Figure 2 . Kaplan–Meier curves for 5-year survival for CIMP + and CIMP − cancers. CIMP, CpG island methylator phenotype.

0.00

0.25

0.50

0.75

1.00

Pro

port

ion

surv

ivin

g

0 20 40 60


CIMP– CIMP+


1193

CO

LON

/SM

ALL

BO

WE

L


CIMP as well as a lack of consensus regarding methylation mark-

ers which most accurately test for CIMP, the number of methylated

markers needed to diagnose CIMP and molecular assay techniques

used to determine the presence of methylation. Our defi nition of

CIMP was based on a panel of six CIMP markers that have been

used to evaluate CIMP status of colon cancer specimens in several

large patient cohorts ( 11,12 ).

Our study is the fi rst to describe the prevalence of CIMP in

interval cancers. Th e prevalence of CIMP was 57 % in interval

cancers compared to 33 % in non-interval cancers, with an overall

prevalence of 41 % in the entire study population. Th e prevalence

of CIMP in the original cohort of colon cancers reported by Toyota

et al. was 50 % ( 10 ). However, in subsequent population based

studies the prevalence of CIMP has ranged from 12 to 29 %

( 11 – 13 ). Th is is similar to prevalence seen in our non-interval

cancer population. In contrast, interval cancers were 2.5 times

more likely to demonstrate CIMP supporting our hypothesis that

tumorigenesis in interval cancers follows a diff erent molecular

pathway than non-interval caners. Consistent with work by others

we also found that CIMP + cancers were more likely to be located

proximally ( 11,13,14 ).

We also confi rmed our previous associations of interval can-

cers with proximal location and MSI, and associations of MSI

tumors with interval cancers, proximal location, poor grade and

mucinous histology ( 9 ). It is important to note that in our study,

CIMP + and MSI are independently associated with interval can-

cers, suggesting that they may operate through diff erent pathways

( 11,14 ). In our cohort, 13 of 26 subjects with MSI lacked CIMP.

As stated previously, sporadic microsatellite unstable tumors

show a strong correlation with CIMP. It has been suggested that

microsatellite unstable tumors lacking CIMP (MSI + / CIMP − )

represent Lynch syndrome. We did not perform germline testing

for mismatch repair genes for any of the subjects, and although

the possibility cannot be entirely ruled out, the following fac-

tors make it unlikely: older age of our study population, the

higher prevalence of MSI + / CIMP − compared to what would be

expected for prevalence of Lynch Syndrome, negative testing of

normal colonic tissue for MSI + and routine questioning at time

of colonoscopy regarding high risk family history. Our fi ndings

support an independent role of CIMP and MSI in the pathogen-

esis of interval cancer.

We found no diff erence in survival between interval and

non-interval cancers. Similarly, there was no diff erence in sur-

vival between CIMP + and CIMP − cancers. Th ere are confl ict-

ing reports on the eff ect of CIMP on patient survival with most

authors reporting no survival benefi t or worse patient survival in

association with CIMP however in a report CIMP was associated

with improved patient survival ( 8,17 – 21 ). Th ese discrepancies are

partly due to inconsistencies in CIMP analysis as described above

as well heterogeneity among populations being studied. In addi-

tion, diff erences in survival have been attributed to the presence

of additional molecular changes involving microsatellite insta-

bility and BRAF and K-Ras mutations. Survival is improved in

CIMP + tumors demonstrating microsatellite instability compared

to CIMP + microsatellite stable (MSS) cancers ( 13,16 ). Among

CI 31 – 53 % ) for CIMP − , P = 0.8] or MSI vs. MSS tumors 26 %

(95 % CI 10 – 45 % ) for MSI vs. 46 % (95 % CI 36 – 55 % ) for CIMP − ,

P = 0.06], respectively ( Figures 2 and 3 ).

DISCUSSION Our study is the fi rst eff ort to evaluate the role of the CIMP

pathway in the pathogenesis of interval and non-interval can-

cers. We found that interval cancers were 2.5 times more likely

than non-interval cancers to demonstrate CIMP + , and as

shown previously, 2.7 times more likely to demonstrate MSI ( 9 )

and nearly 2 times more likely to occur in the proximal colon.

MSI and CIMP were independently associated with interval

cancer. Th ere was no diff erence in survival between interval and

non-interval cancers.

Th e CpG island methylator phenotype was fi rst described by

Toyota et al. in 1999, and aft er some initial debate, is now a widely

accepted pathway ( 6,8,10 ). Over the last few years the focus has

shift ed towards determining the role of CIMP in the pathogenesis

of colorectal carcinoma and eff orts have been made to determine

whether CIMP represents a mechanistic pathway in colorectal

carcinogenesis or whether it represents the recognition of random

and stochastic methylation events that occur in the pathogenesis of

colon cancers ( 8,11 ). Several cohort and population based studies

have identifi ed molecular and clinicopathologic features associ-

ated with CIMP ( 11 – 14 ). In these studies CIMP has been shown to

occur in sporadic cancers and has been associated with older age,

female sex, proximal location, mucinous histology, and microsat-

ellite instability, presence of BRAF mutation and K-ras mutations

and wild type p53. Many of the same features are characteristic of

sporadic MSI tumors however even aft er exclusion of MSI tumors,

the association with old age, proximal location, mucinous histol-

ogy, BRAF mutation and K-ras mutation persists ( 8,14,15 ). In

addition, CIMP has also been shown to be inversely related to loss

of heterozygosity ( 12,16 ). Th ese fi ndings add weight to the mech-

anistic role of CIMP in the pathogenesis of colon cancer. Th ere

is however continued debate over the exact role of CIMP which

is further compounded by the lack of uniformity in defi nition of

Figure 3 . Kaplan–Meier curves for 5-year survival for MSI and MSS cancers. MSI, microsatellite instability; MSS, microsatellite stable.

0.00

0.25

0.50

0.75

1.00

Pro

port

ion

surv

ivin

g

0 20 40 60


MSS MSI


1194 C

OLO

N/S

MA

LL B

OW

EL

Arain et al.

CIMP + tumors lacking microsatellite instability (i.e., CIMP + ,

MSS tumors), the presence of BRAF or K-Ras mutation is associ-

ated with decreased survival ( 16,20 ).

Consistent with our prior work, we found a trend toward

lower survival for MSI cancers than MSS cancers, a fi nding that

contradicts other studies ( 8,22,23 ). Th e possible explanations

for these diff erences could be the older population in our study,

with competing causes of mortality, and possible length and lead

time bias.

It has been proposed that the mismatch repair defects associated

with microsatellite unstable tumors lead to a rapid accumulation

of mutations necessary for tumoriogenesis and result in acceler-

ated tumor growth. It is possible that the CIMP pathway may also

play a direct role in rapid tumor growth however to our knowledge

there is no data to support this. Th e precursor lesions of some, if

not most such cancers may be sessile serrated adenomas. Sessile

serrated adenomas are more likely to occur in the proximal colon

and have been shown to demonstrate high levels of hypermethyla-

tion as well as MSI ( 19,24,25 ). Others have shown an association

of CIMP with tubulovillous and villous adenomas compared to

tubular adenomas ( 26 ). While this was beyond the scope of our

study, the role of precursor lesions for these pathways needs to be

studies further.

Our study had several limitations. Both CIMP and MSI are

associated with female gender however our study population

consists of mainly elderly, Caucasian men, which limits gen-

eralizability ( 11,13 ). We are limited by sample size for detailed

comparisons of tumor characteristics, treatment and survival

diff erences. Also due to sample size limitations we are unable to

assess the combined eff ect of microsatellite and CIMP status on

tumor characteristics, stage at diagnosis and survival. Other lim-

itations include cross sectional design and use of archived tissue

with variable quantity and length of storage and retrospective

nature of study with potential missing / incomplete information

on clinical and pathological variables. Particularly, lack of infor-

mation on smoking and germline testing for mismatch repair

genes are signifi cant limitations.

Our fi ndings suggest that interval cancer result from alternate

pathways, namely CIMP and MSI. We hypothesize that some of

these tumors grow rapidly while others may be diffi cult to recog-

nize during colonoscopy possibly due to a sessile or fl at growth

pattern. Our fi ndings need to be confi rmed in other populations,

particularly women, younger individuals and non-Caucasian

races. Future studies should address the role of CIMP in rapid

tumor growth, particularly in conjunction with MSI. Association

of CIMP + and MSI with dietary and lifestyle factors also need to

be explored.

CONFLICT OF INTEREST Guarantor of the article: Aasma Shaukat, MD, MPH.

Specifi c author contributions: Mustafa A. Arain and Aasma Shaukat:

design of study, data collection and analysis, writing of the report,

and approval of the fi nal draft . Mandeep Sawhney: study design,

data analysis and interpretation, writing of the report, and approval

of the fi nal draft . Shehla Sheikh: data collection and analysis and

approval of the fi nal draft . Ruth Anway: data collection and approval

of the fi nal draft . Bharat Th yagarajan: data analysis and approval of

the fi nal draft . John H. Bond: design of study, data analysis, writing

of the report, and approval of fi nal draft .

Financial support: Supported in part by grants from the VA

Minneapolis Center for Epidemiological and Clinical Research

(CECR) # 04S-CRCOE-001 (A.S.) and the Minnesota Veterans

Research Institute (A.S.)

Financial confl icts of interest: None.

Study Highlights

WHAT IS CURRENT KNOWLEDGE 3 Sporadic colorectal cancers developing after complete

colonoscopy (interval cancers) may represent missed lesions or rapidly growing neoplasms, and in comparison with non-interval cancers may identify differences in genetic pathways.

WHAT IS NEW HERE 3 The current study furthers our understanding of molecular

pathways underlying interval cancers. Our fi ndings suggest that interval cancer results from alternate pathways, namely CpG island methylator phenotype and microsatellite instability.

REFERENCES 1 . Levin B , Lieberman DA , McFarland B et al. Screening and surveillance for

the early detection of colorectal cancer and adenomatous polyps, 2008: a joint guideline from the American Cancer Society, the US Multi-Society Task Force on Colorectal Cancer, and the American College of Radiology . Gastroenterology 2008 ; 134 : 1570 – 95 .

2 . Kahi CJ , Rex DK , Imperiale TF . Screening, surveillance, and primary pre-vention for colorectal cancer: a review of the recent literature . Gastroenter-ology 2008 ; 135 : 380 – 99 .

3 . Barclay RL , Vicari JJ , Doughty AS et al. Colonoscopic withdrawal times and adenoma detection during screening colonoscopy . Th e New England Journal of Medicine 2006 ; 355 : 2533 – 41 .

4 . Robertson DJ , Lieberman DA , Winawer SJ et al. Interval cancer aft er total colonoscopy: results from a pooled analysis of eight studies . Gastroenterol-ogy 2008 ; 134 : A-111 – 2 .

5 . Gorski TF , Rosen L , Riether R et al. Colorectal cancer aft er surveillance colonoscopy: false-negative examination or fast growth? Diseases of the Colon and Rectum 1999 ; 42 : 877 – 80 .

6 . Jass JR . Classifi cation of colorectal cancer based on correlation of clinical, morphological and molecular features . Histopathology 2007 ; 50 : 113 – 30 .

7 . Grady WM , Carethers JM . Genomic and epigenetic instability in colorectal cancer pathogenesis . Gastroenterology 2008 ; 135 : 1079 – 99 .

8 . Ogino S , Nosho K , Kirkner GJ et al. CpG island methylator phenotype, microsatellite instability, BRAF mutation and clinical outcome in colon cancer . Gut 2009 ; 58 : 90 – 6 .

9 . Sawhney MS , Farrar WD , Gudiseva S et al. Microsatellite instability in interval colon cancers . Gastroenterology 2006 ; 131 : 1700 – 5 .

10 . Toyota M , Ahuja N , Ohe-Toyota M et al. CpG island methylator phenotype in colorectal cancer . Proceedings of the National Academy of Sciences of the United States of America 1999 ; 96 : 8681 – 6 .

11 . Samowitz WS , Albertsen H , Herrick J et al. Evaluation of a large, popula-tion-based sample supports a CpG island methylator phenotype in colon cancer . Gastroenterology 2005 ; 129 : 837 – 45 .

12 . Goel A , Nagasaka T , Arnold CN et al. Th e CpG island methylator phe-notype and chromosomal instability are inversely correlated in sporadic colorectal cancer . Gastroenterology 2007 ; 132 : 127 – 38 .

13 . Hawkins N , Norrie M , Cheong K et al. CpG island methylation in sporadic colorectal cancers and its relationship to microsatellite instability . Gastroen-terology 2002 ; 122 : 1376 – 87 .

14 . Weisenberger DJ , Siegmund KD , Campan M et al. CpG island methyla-tor phenotype underlies sporadic microsatellite instability and is tightly


1195

CO

LON

/SM

ALL

BO

WE

L


associated with BRAF mutation in colorectal cancer . Nature Genetics 2006 ; 38 : 787 – 93 .

15 . van Rijnsoever M , Grieu F , Elsaleh H et al. Characterisation of color-ectal cancers showing hypermethylation at multiple CpG islands . Gut 2002 ; 51 : 797 – 802 .

16 . Ferracin M , Gafa R , Miotto E et al. Th e methylator phenotype in microsat-ellite stable colorectal cancers is characterized by a distinct gene expression profi le . Th e Journal of Pathology 2008 ; 214 : 594 – 602 .

17 . Ogino S , Meyerhardt JA , Kawasaki T et al. CpG island methylation, response to combination chemotherapy, and patient survival in advanced microsatellite stable colorectal carcinoma . Virchows Arch 2007 ; 450 : 529 – 37 .

18 . Shen L , Catalano PJ , Benson AB III et al. Association between DNA methylation and shortened survival in patients with advanced colorectal cancer treated with 5-fl uorouracil based chemotherapy . Clin Cancer Res 2007 ; 13 : 6093 – 8 .

19 . Ward RL , Cheong K , Ku SL et al. Adverse prognostic eff ect of methylation in colorectal cancer is reversed by microsatellite instability . J Clin Oncol 2003 ; 21 : 3729 – 36 .

20 . Lee S , Cho NY , Choi M et al. Clinicopathological features of CpG island methylator phenotype-positive colorectal cancer and its adverse

prognosis in relation to KRAS/BRAF mutation . Pathology International 2008 ; 58 : 104 – 13 .

21 . Van Rijnsoever M , Elsaleh H , Joseph D et al. CpG island methylator pheno-type is an independent predictor of survival benefi t from 5-fl uorouracil in stage III colorectal cancer . Clin Cancer Res 2003 ; 9 : 2898 – 903 .

22 . Gryfe R , Kim H , Hsieh ET et al. Tumor microsatellite instability and clini-cal outcome in young patients with colorectal cancer . Th e New England Journal of Medicine 2000 ; 342 : 69 – 77 .

23 . Lim SB , Jeong SY , Lee MR et al. Prognostic signifi cance of microsatellite instability in sporadic colorectal cancer . International Journal of Colorectal Disease 2004 ; 19 : 533 – 7 .

24 . Goldstein NS . Serrated pathway and APC (conventional)-type colorectal polyps: molecular-morphologic correlations, genetic pathways, and implications for classifi cation . American Journal of Clinical Pathology 2006 ; 125 : 146 – 53 .

25 . Kambara T , Simms LA , Whitehall VL et al. BRAF mutation is associated with DNA methylation in serrated polyps and cancers of the colorectum . Gut 2004 ; 53 : 1137 – 44 .

26 . Kakar S , Deng G , Cun L et al. CpG island methylation is frequently present in tubulovillous and villous adenomas and correlates with size, site, and villous component . Human Pathology 2008 ; 39 : 30 – 6 .

cimp status of interval colon cancers: another piece to the puzzle

Documents