Colorectal carcinomas with microsatellite instability display a different pattern of target gene mutations according to large bowel site of origin

Colorectal cancer is the third most common neoplasia in the Western world, preceded only by lung cancer in male and breast cancer in female [1]. Approximately 25 to 35% of colorectal cancers are located in the rectum. Multiple differences between cancer of the right and left colon and rectum with regard to epidemiological, clinical behavior, pathological and molecular features suggest that the mechanisms of sporadic colorectal carcinogenesis may differ according to tumor location [2, 3]. A possible explanation for this could be the different embryological origin of the large intestine, as the ascending and two thirds of the transverse colon originate from the midgut and the last third of transverse, descending colon and rectum from the hindgut [4]. Although few studies have separately analyzed rectal cancer they indicate that approximately 80 to 98% of these cancers arise through the chromosomal instability pathway (CIN), presenting high mutational frequency in APC, TP53, and KRAS in addition to numerous chromosome changes [2, 3, 5]. Nonetheless, a few rectal cancers can develop through the microsatellite instability (MSI) pathway [2, 5].

MSI is a widespread instability in coding and noncoding microsatellite sequences, due to mismatch repair (MMR) deficiency [6]. Through the MSI pathway, colorectal cancer progression is accelerated through accumulation of mutations in coding repetitive sequences of target genes with growth-related functions. However, most of the microsatellite mutations observed in MMR-deficient cells are bystander events that do not play a causal role in carcinogenesis. Criteria to clarify what constitutes a true MSI target gene have been proposed, although most studies have relied on mutation frequency data and on functional studies [7, 8]. In colorectal cancer, mutations have been found in a number of genes with key cellular roles, such as growth factor receptors (TGFBR2 and IGF2R), genes involved in apoptosis (BAX), as well as genes relevant for DNA repair (MSH3, MSH6) [6, 9]. MSI frequency reported in rectal cancer is < 10%, but it is unknown whether or not the same target genes are involved [10].

KRAS and BRAF gene mutations are associated with colorectal development through both CIN and MSI pathways. These genes are members of the mitogen activated protein kinase (MAPK) pathway, which regulates cell proliferation, differentiation, senescence and apoptosis [11]. KRAS mutations are present in 30 to 50% of colorectal carcinomas, occurring mainly in codons 12 and 13 [12‐14]. BRAF mutations are present in 3.7 to 21% of colorectal carcinomas, mainly in codon 600 [13, 15‐17]. Only a small number of studies have specifically addressed the mutational frequency of KRAS and BRAF in rectal carcinomas, reporting a mutational frequency of 21 to 46% for KRAS and about 4% for BRAF [12, 13].

We have previously reported the frequency of nuclear (MSI) and mitochondrial instability in a series of rectal and sigmoid carcinomas [18]. In this study, we aimed to further contribute to the understanding of the pathogenetic mechanisms operating in distal colon and rectal cancers compared to those arising elsewhere in the large bowel.

Existing data on MSI-H frequency indicate that it varies from 10 to 20% in sporadic colorectal cancer, but varies from less than 10% in sporadic rectal carcinomas to about 40% in carcinomas from the right-sided colon [10, 17, 18, 29]. About 70% of MSI-H sporadic colorectal cancers present MLH1 promoter hypermethylation [30]. In this study the MLH1 promoter was analyzed in regions D (in the test series) and C (in the MSI-H validation series), which are both strongly associated with MLH1 protein expression [31]. None of the rectal and sigmoid MSI-H carcinomas of both series presented MLH1 promoter hypermethylation, which we confirmed is significantly associated with colon tumors located more proximally (P = 0.004). These data are compatible with the recent observation of Watanabe et al [32] that MLH1 promoter methylation is significantly less common in left than in right MSI-H colorectal cancer. This difference is reflected in distinct gene expression profiles, which could be taken to indicate that left MSI-H colorectal cancer is a pathogenetically different subgroup among MSI-H sporadic carcinomas [32].

Genes with repetitive sequences located in coding regions are prone to mutations in colorectal carcinomas with the MSI-H phenotype. In fact, there is a well established association between an ineffective MMR system and mutations in the target genes that we have studied, which presumably play a relevant role in colorectal carcinogenesis through the MSI pathway [9, 33]. In rectal MSI-H carcinomas of the test series, we detected mutations only in IGF2R and BAX genes, possible indicating that they are also target genes in the MSI pathway in rectal cancer. The TGFBR2 coding (A)10 sequence is among the most frequently mutated (70 to 90%) sites in MSI-H colorectal carcinomas, indicating that alterations in this gene are crucial for the development of MSI neoplasias [9, 34]. Furthermore, the mutational frequency in MSI-H colorectal cancer is 20 to 39% for MSH3 and 30 to 40% for MSH6 [33, 35]. Interestingly, we did not detect any mutation in TGFBR2, MSH3, or MSH6 microsatellite sequences in rectal or sigmoid MSI-H carcinomas in our test series. In order to confirm these findings, we compared our data with those of an independent series of MSI-H carcinomas with origin in each of the large bowel regions. The trend observed in the test series was confirmed in the validation set, and when all MSI-H carcinomas of the two series are grouped together we observed that TGFBR2 and MSH3 mutations were significantly more prevalent in proximal than in distal (sigmoid and rectal) cancers (P = 0.00005 and P = 0.0000005, respectively). This disparity suggests that these genes are not commonly involved in the development of rectal and sigmoid cancer through the MSI pathway or that alternative mechanisms of inactivation exist. Qualitative (type of target gene) and quantitative (number and frequency of altered target genes) differences have been observed regarding MSI-H target genes in different types of cancers. For instance, TGFBR2 mutational frequency is higher in MSI-H colon carcinomas (70 to 90%) than in MSI-H endometrium carcinoma (17 to 19%), suggesting that biological features and functional roles of target genes may differ depending on the tissue of tumor origin [33, 34]. Our data suggest that both the mechanism of MSI-H and its target genes differ in colorectal carcinomas depending on large bowel site of origin.

Based on clinical data, germline mutations in the mismatch repair genes were initially considered unlikely. However, since none of the MSI-H rectal and sigmoid carcinomas presented MLH1 promoter methylation, we performed immunohistochemical staining to determine the expression of the MMR proteins MLH1, MSH2, MSH6 and PMS2. Two MSI-H rectal carcinomas showed absence of MSH2/MSH6 proteins and subsequent mutation analysis demonstrated that one of the patients presented a germline mutation in MSH2 exon 3. We did not detect any mutation either in the MSH2 or MSH6 genes in the second patient, although it may exist in the promoter or intronic regions not probed in this investigation. These two cases presented also alterations in the BAX and IGF2R genes in their carcinomas (Table 1), which are compatible with a constitutional MMR deficiency that leads to an acquired genetic instability. The immunohistochemical expression of all four MMR proteins in the remaining four cases does not necessarily imply normal DNA mismatch repair function, as missense mutations in MSH2, MLH1 or MSH6 genes may give rise to normal protein levels but abnormal function [36]. Although we did not perform MMR mutation analysis in the validation series, the patient age composition as a whole does not fit a Lynch syndrome profile, as the mean age in the validation series is 68 years, with only three cases unmethylated at the MLH1 promoter being of young age (33-41 years old), all rectal or sigmoid. Additionally, de novo germline mutations in other MMR components cannot be ruled out for neither the test (one rectal and the three sigmoid MSI-H tumors with normal MLH1, MSH2, MSH6, and PMS2 immunohistochemical staining) nor the validation series. As mentioned above, TGFBR2, MSH3 and MSH6 microsatellite sequences present high mutational rates in right MSI-H colorectal cancer, demonstrating that alterations in these genes are important for the development of MSI neoplasias [9, 34]. The fact that we detected significantly less mutations in these three genes in distal MSI-H carcinomas, even in a rectal carcinoma arising in an individual with a MMR constitutional deficiency, could indicate that these genes are not essential for the cancer development in the sigmoid and rectum. It would be interesting to examine for location differences of other genes showing a high mutation frequency in MSI-H tumors such as AC1, ACVR2A, HT001, MRE11A, PTHLH, and TAF1B, which all carry mononucleotide repeats in the coding region and show a mutation frequency of ~70% or higher. However, the relevance of these genes in rectal tumorigenesis remains to be clarified [35].

Several studies have demonstrated the relevance of the MAPK signaling pathway in colorectal cancer, particularly involving alterations in the proto-oncogenes KRAS and BRAF. Deregulation of this pathway can result in apoptosis inhibition and uncontrolled cell proliferation [11‐14]. When considering the MSI-H carcinomas of the combined series, BRAF exon 15 mutations were detected in 17% of the rectal/sigmoid carcinomas and in 43% of those located elsewhere in the colon. Although no statistically significant association was found with tumor location (P = 0.249), BRAF mutations were observed more frequently in proximal than in distal carcinomas, which is in agreement with previous studies reporting that BRAF mutations occur more frequently in carcinomas arising in the right colon (17.8% versus 3.6% in rectal carcinomas) [13, 17]. Several articles found an association between BRAF mutations (namely V600E) and the MSI-H phenotype caused by MLH1 gene promoter hypermethylation (frequently in the right colon), but the sigmoid and rectal carcinoma in our test and validation series, respectively, with BRAF mutation were MSI-H and did not present MLH1 promoter hypermethylation [37, 38]. On the other hand, when considering the total series of MSI-H carcinomas, KRAS mutation frequency was similar for proximal (19%) and distal (25%) carcinomas and inferior to the one observed in microsatellite stable tumors, as previously described [13, 14, 24]. These findings demonstrate that the MAPK pathway involvement in rectal and sigmoid cancer occurs preferentially by KRAS activation.

Our findings in two independent series from different countries indicate that the pattern of genetic changes involved in rectal and sigmoid carcinogenesis is partially different from that observed elsewhere in the colon. As the mutational spectrum of the MSI pathway appears to differ, the carcinogenic mechanisms may also be distinct between MSI-H carcinomas of the recto-sigmoid and those localized more proximally. Further investigation is warranted to determine the mechanism of MSI-H and its target genes in distal colorectal carcinogenesis both in patients with sporadic or hereditary disease.