BRAF mutation, CpG island methylator phenotype and microsatellite instability occur more frequently and concordantly in mucinous than non-mucinous colorectal cancer
Article first published online: 27 DEC 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 118, Issue 11, pages 2765–2771, 1 June 2006
How to Cite
Tanaka, H., Deng, G., Matsuzaki, K., Kakar, S., Kim, G. E., Miura, S., Sleisenger, M. H. and Kim, Y. S. (2006), BRAF mutation, CpG island methylator phenotype and microsatellite instability occur more frequently and concordantly in mucinous than non-mucinous colorectal cancer. Int. J. Cancer, 118: 2765–2771. doi: 10.1002/ijc.21701
- Issue published online: 14 MAR 2006
- Article first published online: 27 DEC 2005
- Manuscript Accepted: 19 OCT 2005
- Manuscript Received: 27 APR 2005
- Department of Veterans Affairs Medical Research Service
- The Theodore Betz Foundation Grant
- mucinous colorectal cancer;
- microsatellite instability;
- CpG island methylator phenotype;
- BRAF mutation;
- KRAS mutation
Mucinous colorectal cancer (CRC) has been reported to have distinct clinicopathological and genetic characteristics. However, the incidence and the relationship among microsatellite instability (MSI), CpG island methylator phenotype (CIMP) and BRAF and KRAS mutations in mucinous and non-mucinous CRC are not known. Activating mutations of BRAF and KRAS and their relationship with MSI and CIMP were examined in 83 sporadic CRC specimens (26 mucinous and 57 non-mucinous CRC). MSI, CIMP, BRAF and KRAS mutation were observed in 17, 24, 25 and 36% of the tumors, respectively. BRAF mutation was highly correlated with MSI (p < 0.001) and CIMP (p < 0.001). A higher incidence of MSI (27% vs. 12%), CIMP (38% vs. 18%, p < 0.05) and BRAF mutation (46% vs. 16%, p < 0.01) was observed in mucinous CRC. KRAS mutation (27% vs. 40%) was observed more frequently in non-mucinous CRC. Significantly higher percentages of mucinous CRC (54%, p < 0.05) had MSI or CIMP or BRAF mutations. Concordant occurrence of 2 or more of these alterations was observed in 39% of mucinous CRC and only 11% of non-mucinous CRC (p < 0.01). The more frequent occurrence and closer association among MSI, CIMP and BRAF mutation in mucinous CRC observed in our study further supports the idea that its pathogenesis may involve distinct genetic and epigenetic changes. © 2005 Wiley-Liss, Inc.
Colorectal cancer (CRC) develops as a result of progressive accumulation of genetic and epigenetic alterations.1 There are at least 2 major genetic instability pathways involved in colorectal carcinogenesis, chromosomal instability and microsatellite instability (MSI).2, 3 The chromosomal instability pathway is found in about 80% of CRCs.4 This pathway involves chromosomal aberrations such as loss of heterozygosity of 5q, 17p and 18q, with inactivation of APC, p53 and DCC genes.2 MSI, a second form of genetic instability, is found in most cases of hereditary non-polyposis colorectal cancer (HNPCC) and in about 15% of sporadic CRCs.5, 6, 7 This pathway involves inactivation of DNA mismatch repair genes followed by mutations in mononuclear tracts in the coding region of genes implicated in tumor progression such as TGFβRII and BAX, which leads to uncontrolled growth and decreased apoptosis.8, 9, 10 Chromosomal aberrations are rare in microsatellite unstable tumors.2
Recently, additional pathways involving epigenetic abnormalities have been described.11, 12 These include CpG island methylator phenotype (CIMP) with hypermethylation of CpG islands in the promoter regions of multiple genes such as p16, p14, MGMT and hMLH1, causing transcriptional inactivation of these genes.13 CIMP is found in 20–32% of sporadic CRC and has been frequently associated with MSI through methylation of hMLH1 promoter region.12, 14, 15, 16
The RAS/RAF/MAP kinase cascade is an important pathway that mediates the cellular response to extracellular signals, which regulate cell growth, differentiation and apoptosis.17 Activating mutations of KRAS, a member of RAS family, are present in 30–40% of cancers.18 In 5–15% of sporadic CRCs, activating mutations of BRAF, a member of the RAF gene family, has been reported to occur and is frequently associated with MSI with methylated hMLH1 promoter region.19, 20, 21 However, HNPCC cases carrying germline mutation in hMSH2 or hMLH1 do not exhibit BRAF mutations.22 Recent studies indicate that BRAF mutation is highly associated with both MSI and CIMP in sporadic CRCs.23, 24
Mucinous CRC, a subtype of CRC, is characterized by the abundant production of extracellular mucins and accounts for 5–15% of all CRCs.25, 26 Clinicopathological studies indicate that mucinous CRC often presents at an advanced stage and is more likely to invade the adjacent viscera.27 There is also more extensive lymph node involvement beyond the pericolonic region and more frequent peritoneal dissemination and recurrence.28 Mucinous cancers are therefore frequently less readily resectible and often carry a worse prognosis. In addition, a higher frequency of MSI29, 30 and CIMP,12, 29 and BRAF mutation23, 29 and a lower frequency of p53 mutation,27, 29, 31 have been reported to occur in mucinous CRC as compared with non-mucinous CRC, indicating that mucinous CRC may have distinct molecular pathways of pathogenesis.31
We have recently observed higher frequency of MSI, CIMP and BRAF mutation and lower frequency of inactivation of APC and p53 and KRAS mutation in mucinous CRC when compared with non-mucinous CRC.29 However, neither the relationship among these molecular alterations nor their relationship to demographic and clinicopathological characteristics of the tumors was examined. Therefore, in the present study, we compared the incidence and the association of MSI, CIMP and BRAF and KRAS mutations in mucinous and non-mucinous CRC. We also examined the relationship among these molecular alterations and their relationship to clinicopathological features such as age, gender, tumor location and Dukes' stage.
Material and methods
Patients and tumor specimens
Clinical data and tumor specimens were obtained from a total of 83 patients with sporadic CRC identified from the surgical pathology files in the Department of Pathology at the San Francisco Veterans Affairs Medical Center and University of California San Francisco. Our study was approved by the Institutional Review Board. All CRC samples were from patients with no known familial history of CRC meeting the criteria of HNPCC or familial adenomatous polyposis. All mucinous CRCs had greater than 50% of tumor volume consisting of mucin. There were 26 mucinous and 57 non-mucinous CRC cases.
Microdissection and DNA extraction
Tumors were microdissected from formalin-fixed, paraffin-embedded 7-μm thick histological sections stained with haematoxylin–eosin using a surgical scalpel under microscopic guidance. Adjacent normal mucosa was microdissected from regions at least 1 cm distant from tumors. For genomic DNA isolation, microdissected samples were incubated overnight at 56°C with 0.5% Tween 20 (Sigma, St. Louis, MO), 100 mM of Tris-HCl buffer (pH 7.6), 1 mM of EDTA and 20 μg of proteinase K (Sigma). Samples were incubated at 56°C overnight. Proteinase K was then inactivated by incubating at 95°C for 10 min, and the extracted DNA was stored at −20°C until use.
To determine the expression of hMLH1 and hMSH2 proteins in tumors, paraffin sections were stained with anti-hMLH1 antibody (1:100; clone G168-15, Pharmigen, Palo Alto, CA) and anti-hMSH2 antibody (1:100; clone FE11, Calbiochem, San Diego, CA). After deparaffinisation, sections were subjected to heat-induced antigen retrieval in 10 mM sodium citrate buffer in a pressure cooker (Biocare Medical, Walnut Creek, CA), pH 6.0 for 20 min. Nonspecific protein binding was blocked by incubating sections with 10% goat serum blocking solution (Zymed Laboratories, South San Francisco, CA) for 10 min. Anti-hMLH1 antibody and anti-hMSH2 antibody were applied separately and incubated at 4°C overnight. Sections were rinsed in PBS followed by incubation of biotinylated secondary antibody (Zymed) for 10 min at room temperature. After a brief rinse, streptavidin–enzyme conjugate (Zymed) was applied and incubated for 10 min. Sections were rinsed, followed by incubation with diaminobenzidine (Zymed) for 2–3 min. After counterstaining with haematoxylin, sections were dehydrated in graded ethanol and cleared in xylene before mounting. Distinct nuclear staining was interpreted as positive staining for both hMLH1 and hMSH2.
The MSI status of each sample was analyzed by comparing the polymerase chain reaction (PCR) patterns at 7 polymorphic repetitive loci of BAT26, D5S-1453, D8S-1130, D11S-1999, D17S-969, D17S-1537 and D18-877. Tissue samples and adjacent normal samples were amplified with these markers. Amplifications were carried out in a PTC-100 thermal cycler (MJ Research, South San Francisco, CA), with HotStarTaq DNA polymerase (Qiagen, Valencia, CA). Thermal cycling included an initial denaturation at 95°C for 15 min, 45 cycles of 94°C for 30 sec, 56°C for 30 sec, 72°C for 30 sec, followed by a final extension at 72°C for 10 min. Amplified PCR product was electrophoresed on 6% acrylamide gels and visualized by ethidium bromide staining. Samples are scored for allelic shifts between tumor and adjacent normal DNA. Tumors were classified as microsatellite stable (MSS) if less than 3 markers had allelic shifts, and MSI if 3 or more markers had allelic shifts.
Bisulfite treatment of DNA, methylation-specific PCRand determination of CIMP status
The methylation status of DNA mismatch repair gene hMLH1, tumor suppressor gene p16, hypermethylated gene in cancer HIC1, a member of ras-associated domain family gene RASSF2, a member of inhibitor family of DNA binding to transcription factor ID4 and 2 CpG islands (MINT1 and MINT31) was determined by methylation-specific PCR. The extracted DNA was treated with sodium bisulfite using EZ DNA Methylation KitsTM™ (Zymo Research, Orange, CA) as recommended by the manufacturer. Methylation status of each gene and locus was determined using 1 μl of bisulfite-treated DNA as template for each PCR reaction and primers specific for methylated and unmethylated alleles separately. Amplification was carried out in a PTC-100 thermal cycler (MJ Research) with PCR cycling conditions optimized for each primer set. The primer sequences for each gene and locus tested and the conditions of the PCR including annealing temperatures and size of products are summarized in Table I. PCR products were electrophoresed on 2% agarose gels and visualized by ethidium bromide staining. CIMP status was classified as CIMP-negative if less than 50% of loci were methylated, and as CIMP-positive if 50% or more of loci were methylated. According to previous articles, the prevalence of CIMP was 19,12 2024 and 32%.15 When we used the definition of “50% or more,” CIMP-positive cases were 24% of CRCs that we examined, a result consistent with the previous reports. The methylation status of hMLH1 was confirmed by comparison with immunohistochemical results.
|Gene and locus||Sequence of forward primer||Sequence of reverse primer||Annealingtemperature (°C)||Length (bp)|
BRAF and KRAS mutation analysis
To detect the BRAF V599E mutation, DNA was amplified by PCR with a primer set covering exon 15 of the BRAF gene.22 This codon was chosen because almost all BRAF mutations in CRC occur here.19, 22 For PCR, 1 primer was designed to contain the mutated sequence at the 3′ end. PCR conditions were 94°C for 10 min, 45 cycles of 94°C for 30 sec, 64°C for 30 sec and 72°C for 30 sec, followed by a final extension at 72°C for 10 min. PCR products were visualized by ethidium bromide following electrophoresis on 2% agarose gels. To detect the KRAS mutations, DNA was amplified with a primer set covering codons 12 and 13 of the KRAS gene.22 PCR conditions were the same as those used to detect BRAF mutations.
Comparisons of categorical variables including gender, tumor location, Dukes' stage, MSI, CIMP status and mutations of BRAF and KRAS were made using χ2-test or Fisher's exact test as appropriate. Student's t test was used for comparisons of the mean age of the patients. Mann-Whitney's U test was used for comparisons of Dukes' stage. For all of the analysis, p < 0.05 was regarded as statistically significant.
MSI and CIMP status
The prevalence of MSI in this set of sporadic CRC cases was 17% (14 of 83 cases), and that of MSS was 83% (69 of 83 cases, Table II). The MSI status was associated with older age (76.8 ± 10.1 vs. 66.9 ± 13.5, p < 0.05) and right-sided location (34%, 12 of 35 cases, p < 0.01). There was no significant association between MSI status, gender and Dukes' stage.
|Features||Total||MSS||MSI||P||CIMP(−)||CIMP(+)||p||BRAF W T||BRAF Mut||p||KRAS WT||KRAS Mut||p|
|Number of cases||83||69 (83)||14 (17)||63 (76)||20 (24)||62 (75)||21 (25)||53 (64)||30 (36)|
|Age||68.6 ± 13.4||66.9 ± 13.5||76.8 ± 10.1||<0.05||67.5 ± 14.1||71.9 ± 11.1||ns||66.9 ± 13.8||73.5 ± 11.2||0.052||69.1 ± 12.7||67.6 ± 14.8||ns|
|Male||53||43 (81)||10 (19)||ns||41 (77)||12 (23)||ns||39 (74)||14 (26)||ns||35 (66)||18 (34)||ns|
|Female||30||26 (87)||4 (13)||22 (73)||8 (27)||23 (77)||7 (23)||18 (60)||12 (40)|
|Location of tumor|
|Right||35||23 (66)||12 (34)||<0.01||20 (57)||15 (43)||<0.01||20 (57)||15 (43)||<0.01||24 (69)||11 (31)||ns|
|Left||48||46 (96)||2 (4)||43 (90)||5 (10)||42 (87)||6 (13)||29 (60)||19 (40)|
|A||9||8 (89)||1 (11)||ns||7 (78)||2 (22)||ns||9 (100)||0 (0)||ns||5 (56)||4 (44)||ns|
|B||39||32 (82)||7 (18)||30 (77)||9 (23)||30 (77)||9 (23)||22 (56)||17 (44)|
|C||23||18 (78)||5 (22)||16 (70)||7 (30)||13 (57)||10 (43)||17 (74)||6 (26)|
|D||12||11 (92)||1 (8)||10 (83)||2 (17)||10 (83)||2 (17)||9 (75)||3 (25)|
High frequency CpG island methylation (CIMP positivity) was present in 24% (20 of 83 cases) of sporadic CRC, while CIMP negativity was observed in 76% (63 of 83 cases, Table II). CIMP positive status was associated with right-sided location (43%, 15 of 35 cases, p < 0.01). There was no significant difference between CIMP status and gender and Dukes' stage.
In the 14 cases of MSI, 11 cases (79%) were also CIMP positive, while only 9 of 69 MSS cases (13%) were CIMP positive, indicating a close association between CIMP and MSI status (p < 0.001, Table III).
|Number of cases||83||63 (76)||20 (24)|
|Unstable (MSI)||14||3 (21)||11 (79)|
|Stable (MSS)||69||60 (87)||9 (13)|
Immunohistochemistry of hMLH1 and hMSH2
All samples with hMLH1 promoter methylation showed negative staining with anti-hMLH1 antibody, while all samples without hMLH1 promoter methylation showed positive staining. This result indicates that in sporadic CRC, negative expression of hMLH1 is caused by methylation of the hMLH1 gene promoter, not by mutation.
All samples examined stained positively with anti-hMSH2 antibody, indicating that our cohort likely did not include HNPCC cases.
BRAF and KRAS mutation, and correlation with MSI and CIMP
BRAF mutation was present in 21 of 83 cases (25%), and KRAS mutation was observed in 30 of 83 cases (36%, Table II). BRAF mutation showed significant correlation with right-sided location (43%, 15 of 35, p < 0.01), while KRAS mutation did not show correlation with any of the clinicopathological features examined. Among 21 cases with BRAF mutation, 12 cases (57%) showed MSI (Table IV), while only 2 cases (3%) showed MSI (p < 0.001) among 62 cases with wild type BRAF. A total of 12 out of 21 cases (57%) with BRAF mutation were CIMP positive, whereas only 8 of 62 cases with wild type BRAF (13%) exhibited CIMP positivity (p < 0.001). There was no correlation between KRAS mutation and MSI or CIMP status. Only 1 case out of 21 with BRAF mutation also had a KRAS mutation, indicating that BRAF and KRAS mutations are mutually exclusive (p < 0.001, Table IV).
|Total||MSS||MSI||CIMP (−)||CIMP (+)||KRAS WT||KRAS Mut|
|Number of cases||83||69 (83)||14 (17)||63 (76)||20 (24)||53 (64)||30 (36)|
|Mutant||21||9 (43)||12 (57)||9 (43)||12 (57)||20 (95)||1 (5)|
|Wild type||62||60 (97)||2 (3)||54 (87)||8 (13)||33 (53)||29 (47)|
Table V shows the relationships between the presence of methylation at individual loci and the status of MSI, CIMP and BRAF mutation of the tumors.
|Genes and loci||Total||MSS||MSI||p||CIMP(−)||CIMP(+)||p||BRAF WT||BRAF Mut||p|
|Number of cases||83||69 (83)||14 (17)||63 (76)||20 (24)||62 (75)||21 (25)|
|hMLH1||12 (14)||0||12 (100)||<0.001||1||11 (92)||<0.001||2||10 (83)||<0.001|
|U||71 (86)||69||2 (3)||62||9 (13)||60||11 (15)|
|p16||21 (25)||13||8 (38)||<0.01||9||12 (57)||<0.001||10||11 (52)||<0.001|
|U||62 (75)||56||6 (10)||54||8 (13)||52||10 (16)|
|HIC1||27 (33)||19||8 (30)||<0.05||13||14 (52)||<0.001||16||11 (41)||<0.05|
|U||56 (67)||50||6 (11)||50||6 (11)||46||10 (18)|
|RASSF2||39 (47)||29||10 (26)||<0.05||22||17 (44)||<0.001||26||13 (33)||ns|
|U||44 (53)||40||4 (9)||41||3 (7)||36||8 (18)|
|ID4||22 (31)||15||7 (32)||<0.05||11||11 (50)||<0.001||13||9 (41)||<0.05|
|U||49 (69)||45||4 (8)||45||4 (8)||40||9 (18)|
|MINT1||25 (30)||14||11 (44)||<0.001||7||18 (72)||<0.001||13||12 (48)||<0.01|
|U||58 (70)||55||3 (5)||56||2 (3)||49||9 (15)|
|MINT31||31 (37)||20||11 (35)||<0.001||16||15 (48)||<0.001||20||11 (35)||0.09|
|U||52 (63)||49||3 (6)||47||5 (10)||42||10 (19)|
The methylation status of all 7 loci had a significant correlation with MSI and CIMP status, whereas methylation at hMLH1, p16, HIC1, ID4 and MINT1 was associated with BRAF mutation. There was no correlation between methylation at any loci and KRAS mutation.
MSI, CIMP and BRAF and KRAS mutations in mucinous vs. non-mucinous CRCs
No significant difference in gender, age and location of tumor was observed between mucinous and non-mucinous CRCs, although there was a trend for mucinous CRCs towards advanced Dukes' stage (p = 0.07, Table VI).
|Features||Total||Mucinous cancer||Non-mucinous cancer||p|
|Number of cases||83||26 (31)||57 (69)|
|Age||68.6 ± 13.4||69.0 ± 14.2||68.4 ± 13.2||ns|
|Male||53||17 (32)||36 (68)||ns|
|Female||30||9 (30)||21 (70)|
|Location of tumor|
|Right||35||11 (31)||24 (69)||ns|
|Left||48||15 (31)||33 (69)|
|A||9||2 (22)||7 (78)||0.07|
|B||39||8 (21)||31 (79)|
|C||23||12 (52)||11 (48)|
|D||12||4 (33)||8 (67)|
In our cohort, 27% (7 of 26 cases) of mucinous CRC were MSI, while only 12% (7 of 57 cases) of non-mucinous CRC were MSI (Table VII). This represents a trend for MSI to occur more frequently in mucinous CRC (p = 0.09). Ten of 26 cases (38%) of mucinous CRC were CIMP positive, while only 10 of 57 (18%) of non-mucinous CRC cases were CIMP positive (p < 0.05). BRAF mutation was observed in 12 of 26 mucinous CRCs (46%), compared with 9 of 57 non-mucinous CRCs (16%, p < 0.01). These results indicate that mucinous CRC has a significantly higher incidence of CIMP and BRAF mutation than does non-mucinous CRC. No significant difference was observed in the incidence of KRAS mutation between the 2 histological types of tumor.
|Type of cancer||Total||MSI||MSS||CIMP(+)||CIMP(−)||BRAF Mut||BRAF WT||KRAS Mut||KRAS WT|
|Mucinous cancer||26||7 (27)||19 (73)||10 (38)||16 (62)||12 (46)||14 (54)||7 (27)||19 (73)|
|Non-mucinous cancer||57||7 (12)||50 (88)||10 (18)||47 (82)||9 (16)||48 (84)||23 (40)||34 (60)|
Table VIII shows the frequency and overlap in MSI, CIMP and BRAF mutation between mucinous and non-mucinous CRCs. In all cases of CRC, 30 of 83 cases (36%) showed one or more alterations, 16 cases (19%) showed 2 or more alterations while 9 cases (11%) showed all 3 alterations concordantly. In mucinous CRC, 14 of 26 cases (54%) showed one or more alterations when compared with 16 of 57 cases of non-mucinous CRC (28%, p < 0.05). Two or more alterations were observed concordantly in 10 of 26 cases (39%) of mucinous CRC when compared with 6 of 57 cases (11%, p < 0.01) of non-mucinous CRC, while the concordant expression of all 3 alterations were observed in 5 of 26 cases (19%) of mucinous CRC, compared with 4 of 57 cases (7%, p = 0.09) of non-mucinous CRC.
|Type of cancer||Total||Mucinous cancer||Non-mucinous cancer||p|
|Number of cases||83||26||57|
|No alteration||53 (64)||12 (46)||41 (72)||–|
|1 or more alterations||30 (36)||14 (54)||16 (28)||<0.05|
|2 or more alterations||16 (19)||10 (39)||6 (11)||<0.01|
|All of 3 alterations||9 (11)||5 (26)||4 (7)||0.09|
It has long been recognized that DNA methylation patterns are grossly disturbed in cancer cells.32 In sporadic CRC, the association between biallelic methylation of the CpG islands of hMLH1 promoter and the development of MSI has been established.33, 34 Recent studies also indicated that CpG islands in certain sites in the genomes are preferentially methylated in tumors and are described as CIMP.14, 15 In addition, CIMP positive tumors have been reported to have a distinct genetic profile with a high frequency of MSI and wild type p53.35, 36, 37
Previously, we have compared the frequency of MSI, CIMP, inactivation of APC and p53, mutations of BRAF and KRAS, in mucinous and non-mucinous CRC.29 However, neither the relationship among these molecular alterations nor their relationship to the demographic and clinicopathological characteristics was examined. In our present study, we investigated the incidence and the relationship among high level methylation status of multiple loci (CIMP), MSI, BRAF and KRAS mutations in sporadic CRCs with mucinous and non-mucinous histology.
We identified CIMP status using 7 loci containing CpG islands. In our previous study,29 we used 4 loci (hMLH1, HIC1, MINT1 and MINT31) for the CIMP analysis. In the present study, we have added 3 new loci (p16, RASSF2 and ID4), based on our recent observation that using 7 of these loci may provide a more accurate CIMP status. There is no consensus on what constitutes positive or negative CIMP status at this time. We arbitrarily defined that CIMP positivity of 50% or more of the loci were methylated, since CIMP positive cases were 24% of CRCs using these criteria, a result consistent with the previous reports of 19,12 2024 and 32%.15 The incidence of 17% MSI, 24% CIMP, 25% BRAF mutation and 36% KRAS mutation in sporadic CRCs were observed. These alterations, except for KRAS mutation, occurred more frequently in the right colon and in an older age group, consistent with other studies.12, 23, 30 A close association of MSI and CIMP positivity was observed (p < 0.001). BRAF mutation also showed a close association with MSI, compared with wild type BRAF (p < 0.001). Only 1 of 21 tumors with BRAF mutation also showed KRAS mutation, indicating that BRAF and KRAS mutation is generally mutually exclusive.20, 38 These results are consistent with previously published studies.19, 22
We also observed that high level methylation of 7 loci was associated with both MSI and BRAF mutations. Recently, BRAF mutation has been reported to be correlated with methylation of multiple genes in sporadic CRC.23 In our study, 83% of sporadic CRCs with methylation in hMLH1 showed BRAF mutations. The frequent occurrence of BRAF mutation in the tumors with hypermethylated hMLH1 has previously been reported.21 Similarly, high levels of methylation of 4 other loci; p16, HIC1, ID4 and MINT1 were correlated with BRAF mutations. However, in our analysis, only about half of cases with BRAF mutations (12 of 21 cases) showed MSI simultaneously (Table IV). Thus, BRAF mutation may not be simply a consequence of defective mismatch repair systems but may be due to a defect in other genes. Interestingly, 36% of sporadic CRC had concurrent alterations of MSI, CIMP and BRAF mutation and 19 and 11% had 2 or 3 of these alterations, respectively. These results indicate that these genetic and epigenetic alterations are not a rare event in CRC and that their occurrence may characterize a subset of CRC in which these alterations play an important role in the pathogenesis.
Recently, BRAF mutation was reported to occur more frequently in poorly differentiated/mucinous CRCs than in well and moderately differentiated CRCs.23 However, it was not clear how many poorly differentiated non-mucinous cancers and mucinous cancers were included in this group. In our study, we observed that the incidence of CIMP and BRAF mutation in mucinous CRC was significantly higher than in non-mucinous CRC. Furthermore, MSI tends to occur more frequently in mucinous CRC. Fifty-four percent of mucinous CRC had one or more alterations in MSI or CIMP or BRAF when compared with 28% of non-mucinous CRC. Moreover, concordant occurrence of these alterations was observed more frequently in mucinous CRC when compared with non-mucinous CRC. The incidence of KRAS mutation in mucinous CRC has been reported to be either higher or lower than that in non-mucinous CRC.27, 39 In our study, KRAS mutation occurred less frequently in mucinous CRC, although the results were not statistically significant.
Recently, CRCs with MSS and CIMP positivity were reported to have a distinctive pathologic phenotype with both similarities to and difference from MSI-H tumors.40 They observed that MSS and CIMP positive CRCs share some characteristics with MSI-H cancers such as right-sided location and poor differentiation. In the present study, 4 mucinous and 5 non-mucinous cancer samples were CIMP positive and MSS. Among these samples, 5 of 9 cases (55%) were located in the right side, and all 5 non-mucinous cancers were well to moderately differentiated, 2 showed cribriform pattern with necrosis, 1 showed prominent tumor infiltrating lymphocytes and 1 showed a minor extracellular mucinous component. Since 18 of 60 cases (30%) of CIMP negative and MSS tumors were located in the right side, our results are consistent with the previous report.40 Although some of these features confirm to the earlier observation,40 the numbers are too low to make a definite statement regarding the comparison of morphologic features between MSI tumor and MSS tumors with positive CIMP. All CIMP positive and MSS tumors showed no hMLH1 methylation. This result indicates that in these tumors, CIMP has not involved in the promoter region of hMLH1 gene, whose methylation causes MSI. These results further illustrate the complexity of CRC tumorigenesis.
In conclusion, these findings suggest that CRC is a heterogeneous disease involving a complex array of genetic and epigenetic aberrations. Tumors with MSI, CIMP and/or BRAF mutation may characterize a subset of CRC with distinct molecular and biological properties and prognosis. Our study also indicates that mucinous CRC exhibits different genetic and epigenetic characteristics with frequent association with MSI, CIMP, and BRAF mutations and suggests that mucinous CRC may arise from alternative pathogenic pathways.
We thank Drs. James Gum and Suzanne Crawley for helpful suggestion and editorial assistance, and we also thank Ian B. Bell and Anh Nguyen for helpful technical support during the course of our study.