Serrated Adenomas Have a Pattern of Genetic Alterations That Distinguishes Them from Other Colorectal Polyps

Colorectal neoplasms result from the sequential accumulation of alterations in genes that regulate cell growth. These alterations include activating point mutations of K-ras (1) and inactivation of several tumor suppressor genes, most notably the APC gene on chromosome 5q21 (2), the p53 gene on 17p13 (3), and one of several candidate tumor suppressor genes on chromosome 18q, most likely DPC4/SMAD4 (4). Different patterns of genetic alterations in tumor suppressor genes, DNA mismatch repair genes, and K-ras have been correlated with distinctive phenotypes of colorectal cancer, precursor neoplasms, and other polypoid lesions in the colon (5, 6).

It is currently appreciated that colorectal neoplasia may evolve through one of several different genetic pathways (7). Not all colonic polyps have the same malignant potential. Adenomatous polyps are the classic precursors of colorectal cancer, but other lesions, such as hyperplastic polyps are thought not to have a propensity for such evolution.

Serrated adenomas are polypoid lesions present in the colon that are characterized by saw-toothed or serrated crypts with dysplasia, and are distinguished from classical adenomas and hyperplastic polyps by their histologic appearance (8). Serrated adenomas were recognized as a distinct entity in 1990, and have also been referred to as mixed hyperplastic adenomatous polyps (9). It has been reported that serrated adenomas make up ∼1% to 2% of all colorectal polyps (10). Serrated adenomas have been reported to have infrequent mutations of APC (11), frequent methylation of the promoters of putative tumor suppressor genes (12), and other genetic alterations (13). There has not been a systematic search for patterns of genomic instability in these lesions, and the relevance of these lesions to a multistep carcinogenesis pathway remains uncertain. In this study, we tested the hypothesis that serrated adenomas might have a unique mutational spectrum compared with other types of lesions in the colon, which could provide insight into the biological basis and clinical potential of these lesions.

A total of 66 small sporadic colorectal polyps, all from different individuals and <10 mm in diameter, were obtained for this study. Eleven lesions were hyperplastic polyps, and 55 were various adenomatous polyps including 27 serrated adenomas, and 28 other types of adenomas (19 simple tubular adenomas, 6 tubulovillous adenomas, and 3 villous adenomas); all were excised endoscopically at the Karolinska Hospital, in Stockholm, Sweden.

DNA was extracted from 10% buffered formalin-fixed, paraffin-embedded tissue sections. Genomic DNA was isolated from the excised microdomain pellets as previously described (5).

The DNA was amplified by two-step PCR. PCR amplification of exon 1 of a K-ras fragment containing codons 12 and 13 was first done by the following primers: forward 5′CGTCCACAAAATGATTCTGAATTAGCTGTATC3′; reverse 5′CCTTATGTGTGACATGTTCTAATATAGTCAC3′. Thirty-five cycles (92°C for 30 seconds and 67°C for 30 seconds) were done followed by a 10-minute extension at 72°C. The initial PCR products were diluted and further amplified with hemi-nested primers: forward 5′AGGCCTGCTGAAAATGAC3′; reverse, the same primer as described above. Thirty-five cycles of amplification (92°C for 25 seconds, 55°C 25 seconds, and 72°C for 25 seconds) were done followed by a 10-minute extension at 72°C. The amplified DNA fragments were then dot-blotted onto nylon filters (Hybond-N, Amersham, Buckinghamshire, United Kingdom) and hybridized with radiolabeled oligomer probes for all possible mutations of codon 12 and the GAC mutation of codon 13.

The DNA was amplified by PCR using 5′ ³²P-end labeled primers using multiple microsatellite markers at microsatellite loci linked to APC on 5q, DPC4/SMAD4 on 18q, p53 on 17p, and the DNA mismatch repair genes hMSH2 on 2p and hMLH1 on 3p as previously described (5). PCR reactions were carried out as previously described, and repeated at least twice to ensure that the results were reproducible in each case.

Assessment of allelic imbalance (also called loss of heterozygosity or LOH) was assigned when a tumor allele showed at least a 50% reduction in the relative intensity of one allele in neoplastic tissue compared with matched normal DNA. Microsatellite instability (MSI) was defined by band shifts at two or more microsatellite loci, which is MSI-high, or MSI-H (14). Band shifts in only one of the markers was defined as MSI-L. Samples with MSI were not considered informative at that locus for the LOH analyses. Similarly, if only one band were produced at PCR in the normal tissue, these samples were not informative for that locus for assessing allelic imbalance/LOH, although these were potentially useful for MSI analysis.

Comparisons between the two groups were done by Fisher's exact test using the statistical software package StatView. P values ≤ 0.05 were considered statistically significant.

Serrated adenomas were characterized by saw-toothed, elongated, and dilated crypts with nuclear atypia. Hyperplastic polyps also showed saw-toothed and dilated crypts and were distinguished from serrated adenomas by the absence of significant nuclear atypia.

K-ras mutations were detected by dot-blot hybridization of PCR products, and confirmed by direct sequencing. The genetic alterations in all 66 colorectal polyps are summarized in Table 1. Tables 2 and 3 provide details of the genetic alterations identified in 27 serrated adenomas and 11 hyperplastic polyps, respectively. As shown in Table 1, serrated adenomas had significantly fewer K-ras mutations (2 of 27; 7%) compared with the other adenomas (10 of 28; 36%; P = 0.011, Fisher's exact test). In contrast, no significant difference was found between lesions with low-grade dysplasia versus high-grade dysplasia. Four of 11 (37%) hyperplastic polyps had K-ras mutations, which is the same as that found in nonserrated adenomas (36%). K-ras mutations were significantly associated with an exophytic appearance of an adenoma (P = 0.029).

Allelic imbalance or LOH at 18q was significantly more frequent in serrated adenomas than in other adenomas (P = 0.046), being found in 7 of 25 (28%) and 2 of 27 (7%) informative serrated adenomas and other adenomas, respectively. LOH at 5q was not found in any serrated adenoma, but was present in 6 of 28 (21%) other adenomas (P = 0.016). LOH at 2p (hMSH2) was found in 3 of 23 informative serrated adenoma, but LOH at 2p was not found in other adenomas. LOH on 3p (hMLH1) was found in 5 of 23 and 3 of 24 informative serrated adenomas and other adenomas, respectively. LOH at the DNA mismatch repair genes was not significantly different between any adenoma types. LOH at 17p (p53) was not found in any informative case, as expected (15). Hyperplastic polyps had LOH at 3p (2 of 10) and 18q (2 of 10), but no LOH at 2p, 5q, and 17p. No significant difference was found in the frequency of LOH on each locus between nuclear grades.

We identified MSI-H in 1 of 27 (4%) of the serrated adenomas, 0 of 28 other adenomas, and 0 of 11 of the hyperplastic polyps. MSI-L was identified in 11 of 27 (41%) of the serrated adenomas, and 11 of 28 (39%) other adenomas. Five of the 11 (45%) hyperplastic polyps showed MSI-L. There was no statistically significant difference in MSI-H or MSI-L occurrence among the subtypes of polyps.

In this study, serrated adenomas were significantly more likely to have allelic imbalance or LOH of 18q, and were significantly less likely to experience allelic imbalance of 5q than other adenomas. Progressive accumulation of inactivated tumor suppressor genes has been consistently identified in colorectal cancer, involving the APC gene at 5q as an early event, a chromosome 18q allele at the DPC4/SMAD4 locus as a progression event, and the p53 gene as the adenoma-to-carcinoma event (16). However, all colon cancers do not always follow these multistep sequence paradigms. Some colon cancers have allelic imbalance at 18q without allelic imbalance of 5q. Serrated adenomas might be the precursors of such of colon cancers. Allelic loss of the DCC locus on 18q was initially reported in 40% to 60% of colorectal cancers, was thought to occur at some time before malignant conversion (16). DPC4/SMAD4, which maps to 18q21, has been identified as a more likely candidate tumor suppressor gene frequently mutated or deleted in colon cancers (17).

The DPC4/SMAD4 protein acts as a factor required in the transforming growth factor-β–mediated signal transduction cascade. Transforming growth factor-β stimulates the production of extracellular matrix, including glycoproteins, fibronectin, and laminin from various cells (18). Inactivation of DPC4/SMAD4 might result in the abnormal accumulation of mucin in cellular cytoplasm because of a disorder of extracellular matrix production by the aberrant transforming growth factor-β signal. One of the speculations for the serrated appearance has been the abnormal accumulation of mucin in cellular cytoplasm (19). These findings suggest that 18q21 allelic imbalance might be an important event in the morphogenesis of serrated adenomas, and inactivation of DPC4/SMAD4 at the 18q locus might be a key genetic alteration in these lesions. Makinen et al. (20) have suggested that mucinous adenocarcinomas are more frequent (9 of 27 cases) when there is an adjacent serrated adenoma.

One group has speculated that there may be a serrated adenoma-carcinoma sequence (13, 21). Their observations suggest that some mucinous adenocarcinomas may evolve through a serrated adenoma pathway. It has been reported that hyperplastic polyps and serrated adenomas of the colorectal show a mixed differentiation pattern, expressing both gastric (MUC1 or MUC5AC) and intestinal (MUC2) mucins (12). pS2 and human gastric mucin are expressed significantly more frequently in both hyperplastic polyps and serrated adenomas than in tubular adenomas or adenocarcinomas. It has also been reported that the serrated adenoma can show some degree of gastric differentiation (22). In future studies, it may be necessary to study inactivation of DPC4/SMAD4 and the expression of gastric mucin phenotypes in mucinous adenocarcinomas of the colon.

In this study, we found that serrated adenomas had significantly less allelic imbalance at 5q, and fewer K-ras mutations, which have been noted to be frequent, early events in the genesis of the adenomatous polyp. The K-ras gene encodes a 21 kDa guanosine triphosphatase protein (p21^ras), which mediates signaling events regulating cell proliferation, and a single mutationally activated p21^ras protein continuously induces cell proliferation as a genetically “dominant” effect. Increased cell growth induced by activating K-ras mutations and inactivation of APC seems to favor an exophytic appearance of colorectal neoplasms (5).

It has been reported that APC mutations are relatively infrequent in serrated adenomas (11), whereas the majority of colorectal neoplasms have inactivating mutations at the APC locus (2). K-ras mutations and 5q allelic imbalance events do not seem to be important in the morphogenesis of serrated adenomas, which indicates a different origin of these lesions compared with other adenomas in the colon.

Hyperplastic polyps are generally regarded as nonneoplastic lesions, however, it has been reported that some of these polyps have hyperproliferative activity (23), overexpression of p53 (24), some form of genomic instability (25), and K-ras mutations (26). In this study, we found that hyperplastic polyps have occasional allelic imbalance events at 18q, but not 5q. Serrated adenomas and hyperplastic polyps have some similarities in their genetic mutational signatures, including frequent 18q allelic imbalance and infrequent 5q allelic imbalance. Serrated adenomas have histologic features that resemble hyperplastic polyps, such as the saw-toothed or serrated crypts. Several reports have noted the histologic and genetic similarities between hyperplastic polyps and serrated adenomas. Our findings suggest that serrated adenomas and hyperplastic polyps have more than morphologic similarities.

Allelic imbalance at 17p was not found in any of the 66 polyps. These polyps were all small (i.e., <10 mm), and removed colonoscopically. It has been reported that allelic imbalance at 17p is frequently found in early colon cancers (5). Taken together, these findings lend additional support to the conclusion that LOH at p53 actually mediates the adenoma to carcinoma transition (15).

In summary, serrated adenomas are significantly more likely to have an LOH event on 18q and are significantly less to experience allelic imbalance at 5q, or have K-ras mutations, compared with other adenomatous polyps in the colon. Serrated adenomas may evolve through a different genetic pathway than other types of adenomas in the colon.

The authors gratefully acknowledge Katsumi Miyai, Department of Pathology at the University of California, San Diego, for critical histologic consultation.