relationship between genetic alterations and prognosis in sporadic colorectal cancer

7

Click here to load reader

Upload: shih-ching-chang

Post on 12-Jun-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

Relationship between genetic alterations and prognosis in sporadic

colorectal cancer

Shih-Ching Chang1,2, Jen-Kou Lin2, Shung Haur Yang1,2, Huann-Sheng Wang2, Anna Fen-Yau Li3

and Chin-Wen Chi4,5*

1Institute of Clinical Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan, Republic of China2Division of Colon & Rectal Surgery, Department of Surgery, Taipei Veterans General Hospital,National Yang-Ming University, Taipei, Taiwan, Republic of China3Department of Pathology, Taipei Veterans General Hospital, National Yang-Ming University, Taipei, Taiwan,Republic of China4Institute of Pharmacology, School of Medicine, National Yang-Ming University, Taipei, Taiwan, Republic of China5Department of Medical Research and Education, Taipei Veterans General Hospital, Taipei, Taiwan, Republic of China

Because chromosomal chromosomal instability (CIN) and micro-satellite instability (MSI) are important genetic alterations in co-lorectal cancers, we classified the sporadic colorectal cancers(CRC) on the status of the CIN and MSI and explored their mole-cular profiles. A total of 213 colorectal tumors were collected foranalysis of DNA ploidy, MSI, loss of heterozygosity (LOH), muta-tion of p53 (exons 5 to 9), Ki-ras (exons 1 and 2) and BRAF(V599E). Relationships between clinicopathological variables andmolecular analyses were analyzed with the v2 test (Yates’ correc-tion). Kaplan–Meier survival curves were compared using log-rank test. Variables with p < 0.1 were entered into the Cox regres-sion hazard model for multivariate analysis. High microsatelliteinstability (MSI-H) existed in 19 tumors (8.9%), which were morelikely to be right-sided (31.6%) with poor differentiation (26.3%).Seventy-one (33.3%) tumors were diploid and 142 (66.7%) wereaneuploid. Mutations in p53, Ki-ras and BRAF were found in45.1%, 41.8% and 4.2% of tumors, respectively. Based on MSI,and CIN, 3 classes were defined: (i) High microsatellite instabilityMSI-H tumors: young age, high carcinoembryonic antigen (CEA)level, right colon, poorly differentiated, mucin production, highBRAF mutation, lower allelic loss and relatively good prognosis;(ii) Microsatellite stability (MSS) diploid tumors: right colon,poorly differentiated, less infiltrative tumor, mucin production,lower allelic loss and low p53, BRAF mutation; (iii) MSS aneu-ploid tumors: more infiltrative invasion, greater allelic loss andhigh p53 mutation. According to multivariate analysis, tumorstage and p53 mutation were significantly associated with diseaseprogression. The MSS diploid and MSS aneuploid CRCs could besubtyped with p53 mutation and had different prognostic outcomeand molecular profiles. The 4-year disease-free survival (DFS) ofpatients with MSS-diploid, wild-type p53 tumors was 67% and sig-nificantly higher than those of patients with MSS-diploid, mutantp53 CRC (30%, p 5 0.003). The same trend was found in patientswith MSS-aneuploid CRC(wild p53 vs. mutant p53, 64% vs. 41%,p 5 0.009). We concluded that CIN, MSI and p53 mutation statusmight be used as a multiple parameter profile for the prognosis ofsporadic colorectal cancer.' 2005 Wiley-Liss, Inc.

Key words: p53; aneuploidy; microsatellite instability; colorectalcancer; prognosis

Colorectal cancer is the third most common cancer followedbehind cervix cancer and breast cancer in women and lung cancerand hepatoma in men, in Taiwan.1 The development of most co-lorectal carcinomas is associated with allelic losses of tumor sup-pressor genes 17p (p53 gene), 5q (adenomatous polyposis coli,APC) and 18q (deleted in colorectal cancer, DCC), as well as amutation-based dysregulation of the Ki-ras proto-oncogene.2,3

With recent progress in molecular carcinogenesis, 2 major mecha-nisms of genomic instability have been proposed for colorectalcancer: chromosomal and microsatellite instability.

Aneuploidy (as measured by flow cytometry) reflects an overallchange in chromosome number and is well-known in colorectalcancer,4,5 although its prognostic value is inconclusive.6,7 Chro-mosomal instability (CIN) (i.e., aneuploidy) arising from chromo-

somal segregation abnormalities can occur early in colorectal car-cinogenesis,8 and is characterized by high frequency of alleliclosses, deletions and mutations in tumor suppressor genes such asAPC and p53.9,10 Colorectal cancers may also have intragenicmutations characterized by microsatellite instability (MSI).11 MSIhas been found in more than 90% of cases of hereditary nonpoly-posis colorectal cancer.12 MSI may result from defective DNAmismatch repair genes, such that misaligned repetitive sequencescaused by slippage of polymerases are not repaired. In sporadiccases, high-frequency MSI (or MSI-H) occurs in �15% oftumors.12,13 Studies indicate that MSI-H is an independent indica-tor of good prognosis.14,15 One hypothesis that explains goodprognosis is that MSI-H tumors seldom demonstrate a loss ormutation of the genes associated with poor prognosis such as p53and DCC.16,17 Recent report showed that activating mutations inBRAF, a member of RAF gene family which encode kinases thatare regulated by Ras and mediate cellular responses to growth sig-nals, was found to be associated with MSI-H cancers.18

Colorectal cancers without CIN or MSI may represent a poten-tially distinctive disease pattern. Although the carcinogenesispathway of microsatellite stable (MSS) diploid tumors has notbeen elucidated, MSS diploid tumors seem to have an aggressivebehavior. 19 However, it is interesting that some published reportsdemonstrated that patients with diploid tumors had relatively goodoutcome.6,7,20

In the present study, we classified colorectal cancers into 3 majorgroups: MSI-H, MSS diploid and MSS aneuploid based on micro-satellite analysis and DNA flow cytometry. To explore the geneticpathways of colorectal cancer carcinogenesis, we evaluated therelationships between molecular alteration in genes (including p53,Ki-ras, BRAF), allelic loss and clinical features or status.

Material and methods

Patients and clinical disease status

A total of 213 patients with sporadic colorectal cancer whounderwent surgery in Taipei Veterans General Hospital from

Grant sponsor: Taipei Veterans General Hospital; Grant number:VGH92-0248; Grant sponsor: National Science Council; Grant number:NSC92-2314-B075-100.*Correspondence to: Department of Medical Research and Education,

Taipei Veterans General Hospital, No 201, Section 2, Shih-Pai Road Tai-pei 11217, Taiwan, Republic of China. Fax:1886-2-28751562.E-mail:Received 1 March 2005; Accepted after revision 15 August 2005DOI 10.1002/ijc.21563Published online 17 October 2005 inWiley InterScience (www.interscience.

wiley.com).

Abbreviations: CEA, carcinoembryonic antigen; CIN, chromosomalinstability; CRC, colorectal cancer; DFS, disease-free survival; DI, DNAindex; HNPCC, hereditary nonpolyposis colorectal cancer; LOH, loss ofheterozygosity; MSI-H, high microsatellite instability; MSS, microsatellitestability; OS, overall survival.

Int. J. Cancer: 118, 1721–1727 (2006)' 2005 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

Page 2: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

January 1999 to December 2000 were included. The study wasapproved by the Institutional Review Board of the Taipei Vet-erans General Hospital, and written informed consent for tis-sue collection was obtained from all patients. Exclusion crite-ria were: preoperative chemoradiotherapy, emergent operativeprocedure, death within 30 postoperative days and evidence ofhereditary nonpolyposis colorectal cancer (Amsterdam criteria)or familial adenomatous polyposis. Clinical information wasrecorded prospectively and stored in a database; this informa-tion included (i) age, sex, personal and family medical historyand (ii) tumor size, location, gross appearance, TNM stage,differentiation and pathological prognostic features. Postopera-tively, patients were monitored every 3 months in the first 2years, and every 6 months afterward. The follow-up protocolsincluded physical examination, digital rectal examination, car-cinoembryonic antigen (CEA), Chest X-ray, abdominal sono-gram and computerized tomography if needed. Proton emis-sion tomography (PET) or magnetic resonance imaging (MRI)are arranged for patients with elevated CEA level but uncer-tain focus of tumor recurrence. The last update of medicalrecord was in December, 2004. The patient population wascomposed of 145 (68.1%) men and 68 (31.9%) women. Themean age at tumor resection was 64.3 6 13.1 years (range:23–86 years; median: 67.6 years). There were 40 right-sidedtumors, 79 left-sided tumors and 94 rectal tumors. Stage distri-bution was: 30 (14.1%) stage I cancers, 73 (34.3%) stage IIcancers, 72 (33.8%) stage III cancers and 38 (17.8%) stage IVcancers. Of the stage III and IV colorectal cancer (CRC)patients, chemotherapy was given to 57 (79%) and 29 (76%)patients, respectively.

Tumor tissue

Tumors were dissected meticulously, with samples collectedfrom the 4 tumor quadrants to explore intratumoral heteroge-neity. The corresponding normal mucosa at least 10 cm awayfrom the primary tumor edge was also collected. Tissue frag-ments were immediately frozen in liquid nitrogen and storedat 270�C. Sections of cancerous and counterpart tissues werereviewed and analyzed by a senior gastrointestinal pathologistblinded to patient outcome. Disease stage was determinedwith the TNM classification of the International UnionAgainst Cancer.21 The pathological factors analyzed includedlymphovascular invasion, invasive tumor pattern, grade of dif-ferentiation, mucin production and intratumoral lymphocyteinfiltration. These pathological features are defined by the Col-lege of American Pathologists consensus statement.22 Highcarcinoembryonic antigen (CEA) level was defined as morethan 5 ng/mL.

Microsatellite instability analysis

High-molecular-weight genomic DNA from each tumor andfrom corresponding normal tissue was purified using the QIAampTissue kit (QIAGEN GmbH, Germany). Yield and purity wasdetermined by electrophoresis on 0.8% agarose gel and spectro-photometry absorbance at 260 nm. According to international cri-teria for determination of MSI,23 5 reference microsatellitemarkers were used: D5S345, D2S123, BAT25, BAT26 andD17S250. Primer sequences were obtained from GenBank(www.gdb.org). Detection of MSI was performed as previouslydescribed.9 Briefly, DNA was amplified using fluorescent poly-merase chain reaction (PCR). PCR products were denatured andanalyzed by electrophoresis on 5% denaturing polyacrylamidegels, and results were analyzed by an ABI 377 automatedsequencer (Applied Biosystems). Tumor samples that exhibiteddifferent allele peaks than the corresponding normal sample(s)were classified as MSI for that particular marker. Samples with 2or more than 2 MSI out of 5 markers were defined as MSI-H;others were microsatellite stable (MSS). Analysis was performedtwice if the results were ambiguous.

Loss of heterozygosity analysis

Another 11 microsatellite markers were used for loss of hetero-zygosity (LOH) analysis: D1S2883, D2S123, D3S1611, D5S345,D7S501, D8S254, D17S396, D18S35, TP53.PCR15, TP53.ALUand MYCL1. Of these markers, D2S123 and D5S345 were usedconcomitantly in LOH and MSI analyses. Primer sequences havebeen previously described.9 The LOH index was calculated as fol-lows: the peak height of 2 alleles in each tumor was divided by thepeak height in normal samples: T1:T2/N1:N2, with T1 and N1 repre-senting the peak heights of the tumor and normal sample, respec-tively, for allele 1, and T2 and N2 representing that for allele 2.Allele loss correlates with an LOH index of �0.67 or �1.5.9

Flow cytometry for ploidy status

DNA ploidy was analyzed by flow cytometry, according to themethod of Jass et al.5 Ploidy was quantitated using the DNA index(DI), i.e., the ratio of the mean fluorescence intensity of the G0/G1peak of the tumor cell population to that of the diploid peak.Specimens were considered as diploid (DI 5 1) if they had a sin-gle G0/G1 peak, or as aneuploid (DI „ 1) when they had 2 or morediscrete peaks with an abnormal G0/G1 peak containing at least20% of the total sample nuclei and having a corresponding G2Mpeak. Samples with a coefficient of variation (CV) > 8% wereexcluded from further analysis. Tumors with both diploid andaneuploid subpopulations were classified as aneuploid. AverageCV values were 5.9% for tumor and 2.2% for normal colonmucosa.

Primer sequence of p53, Ki-ras and BRAF

Exons 5 to 9 of p53 (for highly conserved regions), exons 1 and2 of Ki-ras and BRAF (V599E) were selectively amplified by PCRin a DNA thermocycler (ABI Prism 3100 Genetic Analyzer; Per-kin Elmer Applied Biosystems). The sequence of primers were asfollows: for exon 1 of Ki-ras : 50-ACGATACAGCTAATTCA-GAATCA-30 and 50-AGAAGTAATCAACTGCATGCA-30; forexon 2 of Ki-ras: 50-TGTCTCCTTTCCACTGCTATTAG-30 and50-AGGTTCACACAGGGCCTG-30; for BRAF(V599E): 50-GTG-ATTTTGGTCTAGCTACAGA-30 and 50-TAACTCAGCAGCAT-CTCAGG-30.The sequence of primers for p53 had been reported in previous

study.24 Each sample was sequenced on both sense and antisensestrands. Each mutation was confirmed by a second sequencingprocedure on new PCR products.

Statistical analysis

The endpoint measurements of the study were the percentage ofdisease-free patients and overall survival (OS) from the date ofsurgery. The distribution of each clinicopathological trait wascompared using the two-tailed Fisher’s exact procedure and the v2

test. Data are expressed as mean 6 standard deviation (SD) andcompared using the Student’s t test. Kaplan–Meier survival curveswere compared using the log-rank test. Statistical significance wasdefined as p < 0.05. Multivariate analysis was carried out usingthe Cox proportional hazard model. Variables with p < 0.1 in theunivariate analysis were entered into the multivariate analysis(SPSS for Windows version 10.0).

Results

Table I shows the results of molecular analysis of our 213 CRCtumors. Fifty-four tumors (25.4%) exhibited MSI in at least 1marker, with the following descending frequencies: 9.6% forD2S123, 9.0% for BAT25, 7.5% for BAT26, 5.8% for D5S345 and4.7% for D17S250. MSI-H (MSI in 2 or more than 2 loci) wasfound in 8.9% (19/213) tumors. DNA flow cytometry analysesrevealed that 33.3% (71/213) tumors were diploid and 66.7%(142/213) were aneuploid.

Mutations in p53 were found in 45.1% (96/213) of the tumors,most frequently in exon 7 (40/96 or 41.6%), followed by exon 5

1722 CHANG ET AL.

Page 3: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

(25/96 or 26.0%), exon 8 (18/96 or 18.8%) and exon 6 (13/96 or13.5%) with only 1 mutation in exon 9 (1%). The frequency ofmutations was significantly higher in late stage than in early stagecancers (mutation frequency 5 26.7% in stage I tumors vs. 60.5%in stage IV tumors) (p 5 0.001). Other factors including age, sex,tumor location, lymphovascular invasion, invasive tumor patternand grade of differentiation were not associated with p53 mutation(p > 0.05).

Overall 41.8% (89/213) of the tumors had a Ki-ras mutation,with the highest frequency found in codon 12 (71/89 or 79.8%),followed by codon 13 (14/89 or 15.7%) and codons 19 and 61 (2/89 or 2.2%). Tumors with Ki-ras mutations had significantlyhigher frequencies of mucin production than tumors without Ki-ras mutations (18% vs. 6.5%, respectively). Other variables werenot associated with Ki-ras mutation.

BRAF(V599E) mutations were detected in 9 of 213 CRC (4.23%)and were not associated with any clinicopathological factors.

Markers with detectable heterozygous alleles were consideredinformative for LOH analysis. Overall, 79.8% (170/213) tumorsexhibited LOH in at least 1 microsatellite marker. The highest fre-quency of LOH on a tumor suppressor gene was found forTP53.ALU (65%), followed by DCC (64.3%), D8S254 (51.7%)and APC (47.8%). The mean number of LOH in stage I tumorswas 1.68 (24.8%), increasing to 2.88 (48.0%) for stage IV tumors.Tumors with p53 mutations had a significantly higher mean LOHnumber than tumors without p53 mutation (2.93/47% vs. 1.75/30%). Similar mean numbers of LOH were found between tumorswith and without Ki-ras mutation (2.13/34.7% vs. 2.39/41.2%).

There were 8.9% (19/213) MSI-H tumors. The clinicopatholog-ical difference between MSI-H and MSS tumors is shown in TableII. Patients with MSI-H tumors had the features of younger age

(57.0 years), more right-sided location (31.6%), higher frequencyof poor differentiation (26.8%), more mucin production (31.6%)and more elevated CEA level (78.9%) than those of MSS tumors.On the molecular analysis, MSI-H CRC had higher frequency ofBRAF(V599E) mutation (36.8%) and lower allelic loss (1.6 6 1.0)than those of MSS tumors. The highest frequency of instablilitymarker was BAT26 (14/19) followed by BAT25 (8/19).

Based on the status of ploidy, MSS tumors could be classifiedas MSS-aneuploid (138/213) and MSS-diploid (56/213). As shownin Table III, the MSS-diploid tumors had the features of moreright-sided location (26.8%), higher frequency of poor differentia-tion (14.3%), more mucin production (21.4%) than those of MSS-aneuploid tumors. The MSS-aneuploid tumors had higher fre-quency of p53 mutation (58%) and more allelic loss (2.5 6 1.5)than those of MSS-diploid tumors.

Within a median of 48 (4–77) months follow-up, 92 (43.2%)patients had developed recurrent or metastatic disease, with 60 all-cause deaths. The distribution of disease progression was to theliver (47/92), the lung (28/92), the peritoneum (10/92), the localregion (8/92) and other areas (13/92). Table IV shows thatadvanced stage, high CEA level and p53 mutation were signifi-cantly associated with disease progression and poorer 4-year OS.Patients with p53 mutations had 4-year disease-free survival(DFS) and 4-year OS of 43% and 62%, respectively, whilepatients whose tumors had no p53 mutation had DFS and OS ratesof 67% and 80%, respectively. Patients with MSI-H tumors hadhigher 4-year DFS (78%) than those with MSS tumors (54%), butit did not reach statistical significance. Table V showed that tumorstage was the most significant factor associated with disease pro-gression and survival, followed by p53 mutation, grade of differ-entiation and high CEA level. For the classification of instability,CIN and MSI were not independent prognostic factors.

Based on correlations observed in our study and classificationsdescribed in previous reports,25–27 we used 3 variables: MSI,aneuploidy and p53 to separate tumors into different groups: MSI-H tumors, MSS-diploid tumors with wild-type p53, MSS-diploidtumors with mutant p53, MSS-aneuploid tumors with wild-typep53 and MSS-aneuploid tumors with mutant p53 (Figure 1). Asshown in Figure 2, the 4-year DFS of patients with MSS-diploid,wild-type p53 tumors was 67% and significantly higher than thoseof patients with MSS-diploid, mutant p53 CRC (30%, p 5 0.003).The same trend was found in patients with MSS-aneuploid CRC(wild type p53 vs. mutant p53, 64% vs. 41%, p 5 0.009).

TABLE II – COMPARISON OF CLINICOPATHOLOGICAL PARAMETERS BETWEEN MSI–H1

AND MSS2 COLORECTAL CANCERS

MSI–H MSS p value3

Age (years, mean 6 SD) 57.06 12.6 65.36 12.9 0.008Gender (male, n) 11 (57.9)4 134 (69.1) 0.319Location (n)Right 6 (31.6) 34 (17.5) 0.035Left 2 (10.5) 77 (39.7)Rectum 11 (57.9) 83 (42.8)

CEA level (>5 ng/ml, n) 15 (78.9) 97 (50) 0.013Grade of differentiation (poor, n) 5 (26.3) 14 (7.2) 0.005Lymphovascular invasion (n) 6 (31.6) 44 (22.7) 0.382Invasive pattern of tumor (infiltrative, n) 12 (63.2) 142 (73.2) 0.351Mucin component (>50%, n) 6 (31.6) 23 (11.9) 0.017TNM stage (n)I 4 (21.1) 26 (13.4) 0.573II 8 (42.1) 65 (33.5)III 4 (21.1) 68 (35.1)IV 3 (15.8) 35 (18.0)

LOH number 1.66 1.0 2.36 1.5 <0.001p53 mutation (n) 6 (31.6) 90 (46.4) 0.216Ki–ras mutation (n) 7 (36.8) 82 (2.3) 0.647BRAF mutation (n) 7 (36.8) 2 (1.0) <0.001

1MSI–H; high microsatellite instability N 5 19 (8.9%).–2MSS; microsatellite stability; N 5 194(91.1%).–3v2 test except for age and LOH number where Student’s t test was used.–4Values in parenthe-ses indicate percentages.

TABLE I – PROFILE OF GENETIC CHANGES FOUND IN 213 SPORADICCOLORECTAL CANCERS IN TAIWAN

Type No

MSI–H 19 (8.9)1

Aneuploid 142 (66.7)p53 mutation 96 (45.1)Ki–ras mutation 89 (41.8)BRAF mutation(V599E) 9 (4.2)

1Values in parentheses indicate percentages.

1723MOLECULAR CLASSIFICATION OF COLORECTAL CANCER

Page 4: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

Discussion

Our results document distinct clinicopathological characteris-tics of MSI-H, MSS-diploid and aneuploid sporadic colorectalcancers. MSI-H tumors are characterized by a tendency for right-sided location (31.6%), poor differentiation (26.3%), higher mucinproduction (68.4%), higher CEA production (78.9%) and goodprognosis. These findings are similar to those of previous Westernstudies.28–30 Similar to another report from Southeast Asia,31 ourMSI CRCs had the feature of younger age of onset (57 years). Ourstudies showed that the molecular profiles of the MSI-H CRCwere lower allelic loss, lower frequency of p53 mutation andhigher frequency of BRAF mutation. Of 4 MSI-H tumors withaneuploidy, 3 had p53 mutation. This implies that chromosomeand microsatellite instability are not exclusive, and that p53 mightplay a major role in chromosome instability.

Previous Western study had shown that BRAF mutation wasfrequently found in the sporadic CRC with MSI-H but not inhereditary nonpolyposis colorectal cancer (HNPCC).32 Our seriesshowed that the mutation of BRAF(V599E) was 4.2%, which wasmuch lower than that of previous Western series (10–15%).32,33

Nevertheless, majority of this mutation was observed in theMSI-H CRC (7/19; Table II). In contrast to previous Westernstudies which showed that sporadic MSI-H CRC are mainly char-acterized by a very low frequency mutation in Ki-ras when com-pared with HNPCC and MSS CRCs,32–34 our MSI-H CRCsseemed to share similar Ki-ras mutation frequency between MSI-H and MSS CRCs. These findings are very different from that ofthe Western countries (highly age-related).28 The explanation forsuch difference might be the possibility of lack of sporadic MSI-HCRCs with methylation of MLH1 in our series. Sporadic MSI-Hcancer that is caused by mismatch repair deficiency can be diag-nosed by the alteration of mono-nucleotide repeat microsatellitemarkers with high sensitivity and specificity. In our series, the fre-quency of BAT25 and BAT26 alterations within the subset of MSI-H cancers were 8/19 and 14/19, respectively. In contrast to reportdescribed by Cunningham et al.35 that all but one of the MSI-Htumors with hypermethylation of MLH1 demonstrated MSI atBAT 26, there were 5 MSI-H CRCs without BAT26 alteration inour series. Also, we found a mutually exclusive relationshipexisted between BRAF and K-ras mutations in MSI-H CRCs, andthe majority of Ki-ras mutation in MSI-H CRCs were located

TABLE III – COMPARISON OF CLINICOPATHOLOGICAL PARAMETERS BETWEEN MSS COLORECTAL CANCERSWITH AND WITHOUT CHROMOSOMAL INSTABILITY

Aneuploid1 Diploid2 p value3

Age (years, mean 6 SD) 66.16 12.0 63.36 14.8 0.172Gender (male, n) 96 (69.6)4 38 (67.9) 0.816Location (n)Right 19 (13.8) 15 (26.8) 0.041Left 61 (44.2) 16 (28.6)Rectum 58 (42.0) 25 (44.6)

CEA level (>5 ng/ml, n) 70 (50.7) 27 (48.2) 0.893Grade of differentiation (poor, n) 6 (4.3) 8 (14.3) 0.015Lymphovascular invasion (n) 29 (21.0) 15 (26.8) 0.384Invasive pattern of tumor(infiltrative, n)

109 (79.0) 33 (58.9) 0.004

Mucin component (>50%, n) 11 (8.0) 12 (21.4) 0.009TNM stage (n)I 15 (10.9) 11 (19.6) 0.138II 50 (36.2) 15 (26.8)III 45 (32.6) 23 (41.1)/IV 28 (20.3) 7 (12.5)

LOH number 2.5 6 1.5 1.86 1.5 <0.001p53 mutation (n) 80 (58.0) 10 (17.9) <0.001Ki–ras mutation (n) 56 (40.8) 26 (46.4) 0.455BRAF mutation (n) 1 (<1) 1 (1.8) 0.755

1N 5 138 (71.1%).–2N 5 56 (28.9%).–3v2 test except for age and LOH number where Student’s t testwas used.–4Values in parentheses indicate percentages.

TABLE IV – UNIVARIATE ANALYSIS FOR DISEASE–FREE SURVIVAL(DFS) AND OVERALL SURVIVAL (OS)

Variable N 4–yearDFS (%)

p1 4–yearOS (%)

p1

GenderMale 145 55 0.492 70 0.204Female 68 60 76

LocationRight 40 60 0.059 77 0.183Left 79 47 62Rectum 94 65 73

CEA level2

Low 101 68 <0.001 82 <0.001High 112 43 60

TNM stageI 30 82 <0.001 96 <0.001II 73 80 95III 72 43 67IV 38 10 35

Grade of differentiationWell, Moderate 194 58 0.004 72 0.180Poor 19 28 53

Lymphovascular invasionYes 50 44 0.236 71 0.445No 163 59 72

Invasive tumor patternInfiltrative 154 48 0.481 71 0.847Expansive 59 57 70

Mucin production>50% 29 45 0.206 70 0.906<50% 184 57 72

p53 mutationYes 96 43 <0.001 62 <0.001No 117 67 80

Ki–ras mutationYes 89 51 0.520 68 0.699No 124 59 73

MSI statusMSI–H 19 78 0.071 81 0.226MSS 194 54 72

Chromosomal instabilityAneuploid 138 51 0.284 68 0.176Diploid 56 61 77

1Log–rank test.–2Low CEA level, <5 ng/mL; High CEA level, >5ng/mL.

1724 CHANG ET AL.

Page 5: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

in the codon 13 (GGC- > GAC; 4/7). These findings are similarto previous report,34 which showed that the frequency of Ki-rasmutation in MSI-H CRCs without hMLH1 methylation wassimilar to the frequency observed in HNPCC. It implied thatsome sporadic MSI-H tumors had other mismatch repair dysfunc-tion other than hypermethylation of MLH1 promotor.

Similar to a previous report from Taiwan,35 our series showed thatthe frequency of MSI-H CRC was 8.9% which was lower than pre-vious reports from Western countries.12,13 The possible carcinogene-sis of sporadic CRC with MSI-H phenotype seemed to be associatedwith disruption of the DNA methylation pathway caused by dietarydeficiency or genetic predisposition.36,37 In particular, the enzymemethylenetetrahydrofolate reductase (MTHFR) plays a central role infolate metabolism, regulating the flow of folate groups between 2important biosynthetic pathways including DNA synthesis and DNAmethylation.38 The frequency of TT genotype of MTHFR (C677T)was low in Chinese (7.2%) and Taiwanese (6.9%).39,40 These resultstogether with the observation that adequate serum folate level wasfound in Chinese group39 may contribute to the low frequency ofMSI-H CRC in our series as compared to that of Western countries.

The characteristics of tumors without MSI and CI were inter-esting. The frequency of tumors with chromosome and microsa-tellite stability (MSS-diploid, 26.3%) is also comparable withthat published in a previous report.19 As compared to aneuploidCRCs, relatively more poor differentiation and mucin productionwere found in MSS-diploid tumors and our result agrees wellwith the report of Tang et al.35 The molecular profile of MSS-diploid tumors included a relatively low p53 mutation rate(17.9%), low frequency of allelic loss and very low frequency ofBRAF mutation. An array CGH analysis showed that MSS-diploid tumors had fewer gains and losses at chromosomal levelthan those of MSS-aneuploid tumors, but more than those ofMSI-H tumors.41 The present study identified 2 biologically dis-tinct MSS diploid tumor types that could be classified accordingto the status of p53, and which showed markedly different out-comes. Five percent of tumors were diploid, had stable microsa-tellites and contained p53 mutations and other 21% had no p53mutation. The patients with MSS-diploid tumors with mutant p53had a significantly poorer DFS (30%) than those with wild-typep53 tumors (67%). Most published articles, but not all, indicatethat diploid tumors have a better outcome than aneuploidtumors.5–7,42 The disparity of prognosis in patients with diploidtumors might be due to the fact that a small proportion of theMSS-diploid tumors have a p53 mutation.

The majority of sporadic colon cancers could be categorizedinto tumors with defect in the CIN pathway, which is character-ized by a high frequency of allelic losses, and mutations of tumorsuppressor genes such as p53 and APC.10 In our series, a total of

27.6% of tumors showed aneuploidy without p53 mutation. Theywere primarily located in the left colon and rectum. These patientshad relatively good outcomes compared to patients whose tumorshad aneuploid with p53 mutation. These tumors had a lower DI(1.31) and lower frequency of allelic loss (38%) than aneuploidtumors with p53 mutation (1.62, 49%, p < 0.001). These resultswere consistent with previous reports showing that tumors with ahigher degree of DNA aneuploidy (high DI aneuploidy) wereassociated with p53 mutation, and that near-diploid tumors (lowDI aneuploidy) did not have p53 mutations.43,44 Nearly 40% ofthe tumors in our study were highly aneuploid and had p53 muta-tion. These tumors had the following features: a left-side location,infiltrative tumor invasion, lower mucin production, higher allelicloss and poor prognosis. This type of tumors had higher frequencyof allelic loss at NM23 (22/47), APC (31/50) and DCC (39/48)than those of aneuploid tumors without p53 mutation (NM23: 9/39, p 5 0.022; APC: 17/43, p 5 0.031; DCC: 11/28, p < 0.001,respectively). It is reasonable to assume that multiple genes con-trol the complexity of this process and, thus, that a higher allelicloss should result in a higher metastatic potential.45

Previous studies showed that tumors lacking functional p53 grewaggressively and failed to respond to irradiation and chemotherapeu-tic agents, especially 5FU-base chemotherapy.46,47 We analyzed theinfluence of chemotherapy on the effect of p53 on tumor recurrence.Of the 57 patients with stage III tumors who received chemotherapy,26 of 33 patients with p53 mutation in their tumors developed tumor

TABLE V – MULTIVARIATE ANALYSIS FOR DISEASE FREE SURVIVAL (DFS) AND OVERALL SURVIVAL (OS)OF PATIENTS WITH COLORECTAL CANCER

Variable4–year DFS 4–year OS

HR 95% CI p1 HR 95% CI p1

TNMI, II, III, IV 3.01 1.49–6.06 <0.001 3.28 2.26–4.76 <0.001

p53 mutationYes vs. No 2.87 1.60–4.48 <0.001 2.88 1.91–4.15 <0.001

CEA level2

High vs. Low 2.07 1.04–2.85 0.034 1.85 1.07–3.71 0.021Grade of differentiationPoor vs. well/moderate 2.62 1.29–5.32 <0.001 1.90 1.03–2.25 0.034

MSI statusMSI–H vs. MSS 0.49 0.18–1.38 0.179 0.91 0.35–2.36 0.858

LocationRight, Left, Rectum 0.77 0.58–1.02 0.108 0.85 0.85–1.22 0.702

1High CEA level, >5 ng/mL; Low CEA level, < 5 ng/mL.–2p value results from the hypothesis thatthe hazard ratio (as determined by multivariate binary logistic regression analysis) is 1.0.Abbreviations: HR, hazard ratio; CI, confidence interval; CEA: carcinoembryonic antigen; MSI–H:

high microsatellite instability; MSS: microsatellite stability.

FIGURE 1 – Classification tree for sporadic colorectal cancers basedon microsatellite instability, chromosomal instability and presence(1) or absence (2) of p53 mutation.

1725MOLECULAR CLASSIFICATION OF COLORECTAL CANCER

Page 6: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

recurrence significantly higher than those without p53 mutation(7/22, p < 0.001). Because of small sample size, we could not con-clude that the status of p53 induces chemo-resistance.

In summary, these results support the conclusion that sporadiccolorectal cancer is a heterogeneous disease. Tumors from the 3

different genetic/chromosomal pathways (characterized earlier)have different clinicopathological characteristics, genetic profilesand likely outcome. Future studies elucidating the nature of epige-netic events and gene expression in colorectal cancer should beconsidered.

References

1. Cancer registry annual report, Taiwan. Department of Health, theExecutive Yuan, Republic of china. 2000. (http://crs.cph.ntu.edu.tw/crs_c/annual.html).

2. Fearon ER, Vogelstein B. A genetic model for colorectal tumorigene-sis. Cell 1990;61:759–67.

3. Vogelstein B, Fearon ER, Hamilton SR, Kern SE, Preisinger AC,Leppert M, Nakamura Y, White R, Smits AM, Bos JL. Genetic altera-tions during colorectal-tumor development. N Engl J Med 1988;319:525–32.

4. Verhest A, Kiss R, d’Olne D, Larsimont D, Salmon I, de Launoit Y,Fourneau C, Pasteels JL, Pector JC. Characterization of human color-ectal mucosa, polyps, and cancers by means of computerized morpho-nuclear image analyses. Cancer 1990;65:2047–54.

5. Jass JR, Mukawa K, Goh HS, Love SB, Capellaro D. Clinical impor-tance of DNA content in rectal cancer measured by flow cytometry.J Clin Pathol 1989;42:254–59.

6. Zarbo RJ, Nakhleh RE, Brown RD, Kubus JJ, Ma CK, Mackowiak P.Prognostic significance of DNA ploidy and proliferation in 309 color-

FIGURE 2 – (a) Four-year dis-ease-free survival of the MSS-diploid CRC patients with and with-out p53 mutation was 30% and67%, respectively. (p 5 0.003). (b)Four-year disease-free survival ofthe MSS-aneuploid CRC patientswith and without p53 mutation was41% and 64%, respectively. ( p 50.009).

1726 CHANG ET AL.

Page 7: Relationship between genetic alterations and prognosis in sporadic colorectal cancer

ectal carcinomas as determined by two-color multiparametric DNAflow cytometry. Cancer 1997;79:2073–86.

7. Purdie CA, Piris J. Histopathological grade, mucinous differentiationand DNA ploidy in relation to prognosis in colorectal carcinoma. His-topathology 2000;36:121–6.

8. Lengauer C, Kinzler KW, Vogelstein B. Genetic instability in color-ectal cancers. Nature 1997;386:623–7.

9. Lin JK, Chang SC, Yang YC, Li AF. Loss of heterozygosity and DNAaneuploidy in colorectal adenocarcinoma. Ann Surg Oncol 2003;10:1086–94.

10. Kinzler KW, Vogelstein B. Lessons from hereditary colorectal cancer.Cell 1996;87:159–70.

11. Ionov Y, Peinado MA, Malkhosyan S, Shibata D, Perucho M. Ubiqui-tous somatic mutations in simple repeated sequences reveal a newmechanism for colonic carcinogenesis. Nature 1993;363:558–61.

12. Aaltonen LA, Salovaara R, Kristo P, Canzian F, Hemminki A, Pelto-maki P, Chadwick RB, Kaariainen H, Eskelinen M, Jarvinen H,Mecklin JP, de la Chapelle A. Incidence of hereditary nonpolyposiscolorectal cancer and the feasibility of molecular screening for thedisease. N Eng J Med 1998;338:1481–7.

13. Cunningham JM, Christensen ER, Tester DJ, Kim CY, Roche PC,Burgart LJ, Thibodeau SN. Hypermethylation of the hMLH1 pro-moter in colon cancer with microsatellite instability. Cancer Res1998;58:3455–60.

14. Lothe RA, Peltomaki P, Meling GI, Aaltonen LA, Nystrom-Lahti M,Pylkkanen L, Heimdal K, Andersen TI, Moller P, Rognum TO,Fossa SD, Haldorsen T. Genomic instability in colorectal cancer: rela-tionship to clinicopathological variables and family history. CancerRes 1993;53:5849–52.

15. Halling KC, French AJ, McDonnell SK, Burgart LJ, Schaid DJ, Peter-son BJ, Moon-Tasson L, Mahoney MR, Sargent DJ, O’Connell MJ,Witzig TE, Farr GH, Jr, et al. Microsatellite instability and 8p allelicimbalance in stage B2 and C colorectal cancers. J Natl Cancer Inst1999;91:1295–303.

16. Jen J, Kim H, Piantadosi S, Liu ZF, Levitt RC, Sistonen P, KinzlerKW, Vogelstein B, Hamilton SR. Allelic loss of chromosome 18q andprognosis in colorectal cancer. N Engl J Med. 1994;331:213–21.

17. Kahlenberg MS, Stoler DL, Rodriguez-Bigas MA, Weber TK, Dris-coll DL, Anderson GR, Petrelli NJ. p53 tumor suppressor gene muta-tions predict decreased survival of patients with sporadic colorectalcarcinoma. Cancer 2000;88:1814–19.

18. Rajagopalan H, Bardelli A, Lengauer C, Kinzler KW, Vogelstein B,Velculescu VE. Tumorigenesis: RAF/RAS oncogenes and mismatchrepair status. Nature (Lond) 2002;418:934.

19. Hawkins NJ, Tomlinson I, Meagher A, Ward RL. Microsatellite-sta-ble diploid carcinoma: a biologically distinct and aggressive subset ofsporadic colorectal cancer. Br J Cancer 2001;84:232–6.

20. Lanza G, Gafa R, Santini A, Maestri I, Dubini A, Gilli G, CavazziniL. Prognostic significance of DNA ploidy in patients with stage II andstage III colon carcinoma: a prospective flow cytometric study. Can-cer 1998;82:49–59.

21. Sobin LH, Wittekind C. UICC TNM classification of malignanttumors.5th ed. New York: Wiley-Liss, 1997. 66–9.

22. Compton CC, Fielding LP, Burgart LJ, Conley B, Cooper HS, Hamil-ton SR, Hammond ME, Henson DE, Hutter RV, Nagle RB, NielsenML, Sargent DJ, et al. Prognostic factors in colorectal cancer. Collegeof American Pathologists Consensus Statement 1999. Arch Path LabMed 2000;124:979–94.

23. Boland CR, Thibodeau SN, Hamilton SR, Sidransky D, Eshleman JR,Burt RW, Meltzer SJ, Rodriguez-Bigas MA, Fodde R, Ranzani GN,Srivastava S. A National Cancer Institute Workshop on MicrosatelliteInstability for cancer detection and familial predisposition: develop-ment of international criteria for the determination of microsatelliteinstability in colorectal cancer. Cancer Res 1998;58:5248–57.

24. Chang SC, Lin JK, Lin TC, Liang WY. Genetic alteration of p53, butnot overexpression of intratumoral p53 protein, or serum p53 antibodyis a prognostic factor in sporadic colorectal adenocarcinoma. Int JOncol. 2005;26:65–75.

25. Dutrillaux B. Pathways of chromosome alteration in human epithelialcancers. Adv Cancer Res 1995;67:59–82.

26. Iacopetta B. Are there two sides to colorectal cancer? Int J Cancer 2002;101:403–8.

27. Risques RA, Moreno V, Ribas M, Marcuello E, Capella G, PeinadoMA. Genetic pathways and genome-wide determinants of clinical out-come in colorectal cancer. Cancer Res 2003;63:7206–14.

28. Gafa R, Maestri I, Matteuzzi M, Santini A, Ferretti S, Cavazzini L,Lanza G. Sporadic colorectal adenocarcinomas with high-frequencymicrosatellite instability. Cancer 2000;89:2025–37.

29. Risio M, Reato G, di Celle PF, Fizzotti M, Rossini FP, Foa R. Micro-satellite instability is associated with the histological features of thetumor in nonfamilial colorectal cancer. Cancer Res 1996;56:5470–4.

30. Jass JR, Do KA, Simms LA, Iino H, Wynter C, Pillay SP, Searle J,Radford-Smith G, Young J, Leggett B. Morphology of sporadic colo-rectal cancer with DNA replication errors. Gut 1998;42:673–9.

31. Ko JM, Cheung MH, Kwan MW, Wong CM, Lau KW, Tang CM,Lung ML. Genomic instability and alterations in APC, MCC andDCC in Hong Kong patients with colorectal carcinoma. Int J Cancer1999;84:404–9.

32. Deng G, Bell I, Crawley S, Gum J, Terdiman JP, Allen BA, Truta B,Sleisenger MH, Kim YS. BRAF mutation is frequently present insporadic colorectal cancer with methylated hMLH1, but not in heredi-tary nonpolyposis colorectal cancer. Clin Cancer Res 2004;10:191–5.

33. Davies H, Bignell GR, Cox C, Stephens P, Edkin S, Clegg S, TeagueJ, Woffendin H, Garnett MJ, Bottomley W, Davis N, Dicks E, et al.Mutations of the BRAF gene in human cancer. Nature (Lond) 2002;417:949–54.

34. Oliveira C, Westra JL, Arango D, Ollikainen M, Domingo E, FerreiraA, Velho S, Niessen R, Lagerstedt K, Alhopuro P, Laiho P, Veiga I,et al. Distinct patterns of KRAS mutations in colorectal carcinomasaccording to germline mismatch repair defects and hMLH1 methyla-tion status. Hum Mol Genet 2004;13:2303–11.

35. Cunningham JM, Kim CY, Christensen ER, Tester DJ, Parc Y, Bur-gart LJ, Halling KC, McDonnell SK, Schaid DJ, Walsh Vockley C,Kubly V, Nelson H, et al. The frequency of hereditary defective mis-match repair in a prospective series of unselected colorectal carcino-mas. Am J Hum Genet 2001;69:780–90.

36. Tang R, Changchien CR, Wu MC, Fan CW, Liu KW, Chen JS, ChienHT, Hsieh LL. Colorectal cancer without high microsatellite instabil-ity and chromosomal instability–an alternative genetic pathway tohuman colorectal cancer. Carcinogenesis 2004;25:841–6.

37. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Goldman H,Jessup JM, Kolodner R. Methylation of the hMLH1 promoter corre-lates with lack of expression of hMLH1 in sporadic colon tumors andmismatch repair-defective human tumor cell lines. Cancer Res 1997;57:808–11.

38. Shannon B, Gnanasampanthan S, Beilby J, Iacopetta B. A polymor-phism in the methylenetetrahydrofolate reductase gene predisposes tocolorectal cancers with microsatellite instability. Gut 2002;50:520–4

39. Kim Y-I. Folate and carcinogenesis: Evidence, mechanisms, andimplications. J Nutr Biochem 1999;10:66–88.

40. Kelemen LE, Anand SS, Hegele RA, Stampfer MJ, Rosner B, WillettWC, Montague PA, Lonn E, Vuksan V, Teo KK, Devanesen S, YusufS. Associations of plasma homocysteine and the methylenetetrahydro-folate reductase C677T polymorphism with carotid intima mediathickness among South Asian, Chinese and European Canadians.Atherosclerosis 2004;176:361–70.

41. Hsu LA, Ko YL, Wang SM, Chang CJ, Hsu TS, Chiang CW, Lee YS.The C677T mutation of the methylenetetrahydrofolate reductase geneis not associated with the risk of coronary artery disease or venousthrombosis among Chinese in Taiwan. Hum Hered 2001;51:41–5.

42. Jones AM, Douglas EJ, Halford SE, Fiegler H, Gorman PA, RoylanceRR, Carter NP, Tomlinson IP. Array-CGH analysis of microsatellite-stable, near-diploid bowel cancers and comparison with other types ofcolorectal carcinoma. Oncogene 2005;24:118–29.

43. Visscher DW, Zarbo RJ, Ma CK, Sakr WA, Crissman JD. Flow cyto-metric DNA and clinicopathologic analysis of Dukes’ A and Bcolonic adenocarcinomas: a retrospective study. Mod Pathol 1990;3:709–12.

44. Hamelin R, Laurent-Puig P, Olschwang S, Jego N, Asselain B, Remvi-kos Y, Girodet J, Salmon RJ, Thomas G. Association of p53 mutationswith short survival in colorectal cancer. Gastroenterology 1994;106:42–8.

45. Fidler IJ. 7th Jan Waldenstrom Lecture. The biology of human cancermetastasis. Acta Oncol 1991;30:668–75.

46. Symonds H, Krall L, Remington L, Saenz-Robles M, Lowe S, JacksT, Van Dyke T. p53-dependent apoptosis suppresses tumor growthand progression in vivo. Cell 1994;78:703–11.

47. Lowe SW, Bodis S, McClatchey A, Remington L, Ruley HE, FisherDE, Housman DE, Jacks T. p53 status and the efficacy of cancer ther-apy in vivo. Science 1994;266:807–10.

1727MOLECULAR CLASSIFICATION OF COLORECTAL CANCER